Translational & M olecular I maging I nstitute

6th Annual Symposium

April 22, 2016

New York, NY Table of Contents

Symposium Schedule ………………………………………………….……………………………………………………………………...... ……… 1 Message from the Director ……………………………………………..………………..…………….……..…………………….....…………...... 3 Translational and Molecular Imaging Institute ………………………………………………………………………………………...... … 5 Human Imaging Core ……………………………………………………………………………………………………………...... ……….. 9 Small Animal Imaging Core …………………………………………………………………………………………………...... ………… 13 TMII Research Laboratories ………………………………………………………………………………………...... …………………… 16 Biographies of Invited Speakers ……………………………...…………………………………………………………………………...... …...... 21 Zahi A. Fayad, PhD …………………………………………………………..…………………………………………………………...... ……….. 23 Dennis S. Charney, MD …………………………………………..………………………..……………………………………...... …………... 25 Burton P. Drayer, MD …………………………………………..………………………..……………………………………...... …………... 27 Bruce Fischl, PhD ...... ………………………………………………….…………………………………………..…………...... ….... 29 Julie Price, PhD …....………………………………………………………………………………………………………………………...... 31 Mark Griswold, PhD ………………………………………………………………………………………………………...... ……………... 33 Anna Moore, PhD ………….…………………………………………………………………………………...... ……………………...... 35 Debiao Li, PhD ...... ………………………………………………………………………………………………………...... ……... 37 Abstracts Selected for Oral Presentation ………………………………………………………………………………………………...... 39 Neuroimaging …....……………………………………………………………………….……………………………………………...... 41 Cancer/Body Imaging .………………………………………………………………………………...... …………………………….. 43 Nanomedicine ...... ………………...... ……………………………………………….…………………...... …………….... 45 Cardiovascular Imaging ….……………………………………………………………….…...... ……………………………...... …. 47 Abstracts Selected for Poster Presentation……………………………………………...... …………………………………………… 49 Cancer/Body Imaging ….……………………………………………………………….………………………………...... ……………….. 49 Cardiovascular Imaging ……………………………………………………………….………………………………...... …….…… 69 Nanomedicine ……..……………………………………………………………………….…………………...... ………………………. 99 Neuroimaging …....……………………………………………………………...... ………….……………………...... 115 Sponsors & Contributors ……..………………………………………………………….……………………...... ……………………………...... 151 TMII 2015/16 Highlights ..……………………………………………………………….……………………...... ……………………………...... 157

Davis Auditorium – Leon and Norma Hess Center for Science and Medicine Icahn School of Medicine at Mount Sinai – April 22, 2016

SYMPOSIUM SCHEDULE

7:00am – 8:00am Check-in / Walk-up Registration / Poster Set-up (2nd floor) Refreshments and Light Breakfast

OPENING REMARKS

8:00am – 8:15am Zahi A. Fayad, PhD (Director, Translational & Molecular Imaging Institute)

8:15am – 8:30am Dennis Charney, MD (Dean, Icahn School of Medicine at Mount Sinai)

KEYNOTE ADDRESS – Moderator: Prantik Kundu, PhD

8:30am – 9:30am Bruce Fischl, PhD, (Harvard Medical School; Boston, MA) “Computational analysis of functional, connectional and architectonic properties of the human brain with”

9:30am – 10:00am Break

SESSION I – NEUROIMAGING – Moderator: Priti Balchandani, PhD

10:00am – 10:45am Julie Price, PhD (University of Pittsburg; Pittsburg, PA) “In vivo PET imaging of protein targets in Alzheimer’s disease”

10:45am – 11:00am Alan Seifert, PhD (Icahn School of Medicine at Mount Sinai; New York, NY) “DANTE-EPI for CSF Suppression in Cervical Spinal Cord BOLD fMRI at 7T”

SESSION II – CANCER & BODY IMAGING – Moderator: Bachir Taouli, MD

11:00am – 11:45am Mark Griswold, PhD (Case Western Reserve University; Cleveland, OH) “Rethinking the way we do MRI: Magnetic Resonance Fingerprinting ”

11:45am – 12:00pm Stefanie Hectors, PhD (Icahn School of Medicine at Mount Sinai; New York, NY) “Assessment of tumor heterogeneity in hepatocellular carcinoma using combined DCE-MRI and BOLD measurements”

1

12:00pm – 1:00pm Lunch

POSTER SESSION – Moderators: Sara Lewis PhD, Phil Robson, PhD, Carlos Perez-Medina, PhD, Rafael O’Halloran, PhD, Cheuk Tang, PhD

1:00pm – 2:00pm Poster Viewing

SESSION III – NANOMEDICINE – Moderator: Willem Mulder, PhD

2:00pm – 2:45pm Anna Moore, PhD (Harvard Medical School; Boston, MA) “Image-guided Precision Nanomedicine for Cancer Therapy”

2:45pm – 3:00pm Yiming Zhao, PhD (Icahn School of Medicine at Mount Sinai; New York, NY) “Augmenting drug-carrier compatibility improves tumor nanotherapy efficacy”

3:00pm – 3:30pm Break

SESSION IV – CARDIOVASCULAR IMAGING – Moderator: Philip Robson, PhD

3:30pm – 4:15pm Debiao Li, PhD (Cedars – Sinai; Los Angeles, CA) “MR Coronary Angiography and Vessel Wall Imaging”

4:15pm – 4:30pm Joseph Lerman (National Heart, Lung & Blood Institute (NHLBI); Bethesda, MD) “Lack of improvement in aortic vascular inflammation is associated with an increase in coronary plaque burden in psoriasis”

CLOSING REMARKS & AWARDS

4:30pm – 5:00 pm Burton Drayer, MD (Chair of Radiology, Icahn School of Medicine at Mount Sinai)

2 Message from the Translational and Molecular Imaging Institute Director

I am pleased to welcome you to the 6th Annual Translational and Molecular Imaging Institute (TMII) Symposium.

The TMII symposium continues its mission giving researchers and medical professionals an opportunity to gain insight into the current translational imaging research at Mount Sinai and other institutions in and outside the New York metropolitan area.

This year we are excited to have the opening lecture by Dr. Bruce Fischl, from Harvard/MGH. Dr. Fischl will speak about new and advanced computational methods for the study of the brain and disease. As in previous years we also invited some of the greatest pioneers in imaging science (Mark Griswold from Case Western; Anna Moore from MGH; Julie Price from the University of Pittsburgh and Debiao Li from Cedars Sinai) to talk to you about some of the innovative research in imaging and drug delivery that they’re pursuing.

We received over 50 outstanding abstracts by trainees, faculty and staff from across the NY metro area and beyond. From those we have selected a few exceptional examples, of what we think represent new and exciting developments in the field, to give short oral presentations. We encourage you to attend each of the sessions along with the poster presentations and take advantage of the expertise and enthusiasm these researchers have to offer.

Please take the time to review the program book to learn more about the speakers, TMII, its faculty and the resources it has to offer.

The symposium is being held at the main Mount Sinai campus, located on the famed Museum Mile. We hope that you get a chance to take in these fine art institutions and tour the TMII facilities.

Enjoy the symposium.

Zahi A. Fayad, PhD

TMII Director Mount Sinai Endowed Chair in Medical Imaging and Bioengineering Professor of Radiology and Medicine (Cardiology)

3 4

Translational and Molecular Imaging Institute

5 6

Our mission is centered around development, validation, translation and education of innovative technology in biomedical imaging to address both basic and clinical research problems and therefore improve human health.

The Translational and Molecular Imaging Institute (TMII) is a state of the art research facility housed in ~ 20,000 square feet in the Center of Science and Medicine (CSM). TMII (Director, Zahi A. Fayad, PhD) is comprised of faculty, staff and trainees responsible for coordinating and executing all research projects performed in these facilities. Currently TMII has over 50 members with expertise in all aspects of translational imaging research. The faculty consists of Biomedical and Electrical Engineers and Radiologists who are leading experts in neuroimaging, cardiovascular imaging, body/cancer imaging, and nanomedicine. Highly skilled staff provides a full suite of support services for image acquisition, image analysis, scheduling and performance of the proposed experiments.

Access to the TMII facility is based on a fee for service schedule (https://tmii.mssm.edu/imaging- core/resource-fees/). These user fees are calculated to cover the operating and maintenance costs of the instruments and related Core expenses. These rates are determined and periodically reviewed by the Dean’s Office and adjusted to reflect the actual costs. User fees include technical support for operation of imaging equipment. For internal Mount Sinai users, resource usage time is compiled from the web-based scheduling system and charged directly to your account on a monthly basis. Any questions on the charges should be addressed to the TMII Director. The user fee listed in https://tmii.mssm.edu/imaging-core/resource-fees includes support for:

Study Start-up Study design/ IRB review & consultations Maintenance self-scheduling system Scan Protocol Scan parameter Implementation/ Optimization (HiRes Structural, DTI, EPI ...) Adaptation of C2P sequences to our scanners Task Protocol fMRI Software administration (Eprime, PsychToolBox, PschoPy(thon), Matlab, Cogent) Basic task modifications (Triggering, Physio monitoring control, etc)

Stimulus Equipment Visual (LCD, 3D goggles) Audio (Headphones, ear buds) Eye-tracking equipment (not software control support Physiological Set up (SpO2, Heart rate, Respiration, ECG, GSR, EMG, Skin Temperature) Monitoring

7

Data Handling Long-term online image archive (XNAT) Burn anonymized data to CD for external sites Data Analysis Computer Lab with preloaded analysis software (BV, SPM, FSL, Osirix, Custom In House tools, Matlab) PET isotopes Standard tracers: FDG, NaF

Report Radiological read and report of incidental findings

8

A. HUMAN & LARGE ANIMAL FACILITES - CSM – SC2

1. MR/PET (3T) Siemens mMR.

The 3T MR/PET is a fully integrated and capable of simultaneous whole body PET and MRI scanning. This allows more precisely coregistered functional and structural acquisition while r educing the radiation dose in PET imaging by replacing the CT scans with an MRI scan. True simultaneous acquisition of MR and PET data by the hybrid system merges the highly sensitive PET metabolic information with the highly specific MR anatomical and functional information. The 3T MRI system is a whole body imaging system, capable of routine as well as advanced imaging of all body regions. The PET scanner will be fully integrated into the MR, utilizing state of the art solid-state technology for simultaneous PET imaging during MR image or spectrum acquisition. The 3T MR-PET is designed for the purposes of oncological and neurological diagnostic imaging. The highly integrated nature of these systems provides the capability for full spatial and temporal correlation between both modalities. The maximum gradient amplitude will be approximately 40 mT/m per axis, with a maximum gradient slew rate of about 200 T/m/s per axis. The system’s magnet has an integrated cooling system and active shielding. The shimming capabilities include: Active (with 3 electric and 5 electric nonlinear linear shim channels) and Passive shims for maintaining very high homogeneity and excellent image quality over a wide range of applications. Online shimming is performed in less to then 20 seconds in order to optimize homogeneity. The RF transmit and receive system include: compact, air cooled tube RF amplifier providing 35 kW peak power; integrated electronics with cabinet water cooling; integrated circularly polarized whole body RF coil; up to 32 receive channels. The PET system include: adaptation to a work environment within high magnetic fields including APD and LSO based detector technology; adaptation and optimization of numerous MR components to an integrated PET imaging unit; high-resolution, high-count rate, positron emission tomography (PET) imaging of metabolic and physiologic processes; high quality metabolic and anatomic image registration and fusion for optimal lesion detection and identification within the body; state-of-the-art 3D PET data acquisition and analysis tools; state-of-the-art 3D PET reconstruction, attenuation and scatter correction software. Expected PET performance specifications: spatial resolution: <6.5mm; timing resolution: < 4.5 ns; sensitivity: > 0.5%; axial FOV: > 19 cm ; transaxial FOV: up to 45 cm. The system also supports MR and PET gated scan acquisition;

9

support for list mode acquisition, offline histogramming and reconstruction; special calibration. Alignment and quality control sources including shielding; multimodality workplace; 3D iterative reconstruction.

2. 7T Siemens MR whole body scanner. This is an ultrahigh field 7.0 Tesla actively shielded whole body MRI scanner. The super-conducting magnet is self-shielded, reducing its overall footprint and making it compact and lightweight by 7T standards, weighing 24-tons. The (warm) inner bore of the magnet is 82 cm, which houses the 60 CM inner patient bore. The dimensions of the magnet without covers is approximately 2.5 m in length, 2.6 m in width, and 2.65 m in height. The 5-Gauss line extends slightly further than for a 3T scanner with 5.6 m radial and 7.8 m axial dimension. A whole- body gradient system provides gradient amplitude of up to 70 mT/m per axis, and a maximum slew rate of up to 200 T/m/s. The RF transmit system comes with 8 parallel transmit channels; 8 individually shaped RF pulses can be prescribed simultaneously and independently in amplitude and phase. The multi-nuclei package allows for imaging and spectroscopy at non-proton frequencies, i.e. detection of e.g. 19F, 31P, 7Li, 23Na, 13C, 17O. Our 7T/820AS is configured to accommodate an 8- channel Tx-array and 48-channel Rx receivers. Several coils are currently available such as the 1- channel Tx and 32-channel Rx head coil and the 8-channel Tx and 8-channel Rx head coil.

3. 3T Siemens MAGNETOM Skyra. This is an FDA approved 3 Tesla human MRI scanner. Its wide bore design (173 cm system length with 70 cm) can accommodate subjects with larger body compositions compared to the 60 cm bore of a typical clinical 1.5T & 3T. A newly deigned RF system and coil architecture integrates (Tim 4G) with all digital-in/digital-out technology. The scanner has an actively shielded water-cooled gradient system and zero helium boil-off. Specialized RF distribution increases uniformity in all body regions. Onboard software is available for: neuro, angio, cardiac, body, onco applications. A variety of coils for all body parts and configuration is available.

10

4 PET/CT Siemens Biograph mCT. The PET/CT is equipped with a 40 slices multidetector CT and LSO PET crystals. Utilizing timing information (time-of-flight) between the two PET coincidence events, coupled with high definition resolution recovery, the system provides improved image signal- to-noise which can be used to either enhance image quality and/or reduce acquisition time. With a system timing resolution of 555 ps, image quality is clearer with more defined images and provides distortion-free image quality for the entire of the field of view. Specialized image processing techniques utilizing more accurate point spread functions produces higher quality 3D iterative reconstruction with enhanced contrast and higher resolution. Supported image matrices include 128x128, 200x200, 256x256, 400x400, and 512x512.

5 Multidetector CT (MDCT) Siemens Somatom Definition Flash. This Dual Source CT, uses two X-ray sources and two detectors simultaneously, to cover the entire thorax in less than a second. A 2 meter scan requires only 5 seconds, enhancing the efficiency of perfusion or dynamic vascular imaging and reduction the dose for all scans, resulting, e.g. in dose down to sub-mSv for cardiac imaging. Dual Energy automatically provides a second contrast for without any extra dose. Advance software efficiently manages the reduction in dose allowing for: limited exposure to radiation-sensitive organs and increases tissue contrast with no sacrifice to image quality.

6 (2)1.5T Siemens MR MAGNETOM Aera. Short and open appearance (145 cm system length with 70 cm Open Bore Design) can accommodate subjects with larger body compositions compared to the 60 cm bore of a typical clinical 1.5T & 3T. Newly deigned RF system and coil architecture integrates with all digital-in/digital-out technology, one system use standard gradients (33 mT/m @125 T/m/s) and the second system has advanced gradients (45 mT/m @ 200 T/m/s). Actively shielded water-cooled gradient system with zero helium boil-off. Inline software is specially designed for: neuro, angio, cardiac, body, onco, breast, ortho, pediatric and scientific specialties such as Magnetic Resonance Elastography; a technique that measures the stiffness of tissues by introducing shear waves and imaging their propagation.

7 Siemens ACUSAN S3000 ARFI Ultrasound - The ultrasound system automatically produces an acoustic ‘push’ pulse that generates shear-waves, which propagate into the tissue. Using image-based localization and a proprietary implementation of acoustic radiation force impulse (ARFI) technology, shear wave speed may be quantified, in a precise anatomical region, focused on a region of interest, with a predefined size, provided by the system. Measurement value and depth are also reported, and the results of the elasticity are expressed in m/s. This system provides new method for the evaluation of the elastic properties of tissues is now available in the Cancer/Body Core. Clinical applications of ARFI imaging include: liver fibrosis quantification, breast, colorectal and prostate tumor imaging.

8 Siemens Force CT. The Force is the third iteration of Siemens' dual-source CT design which features two sets of x-ray tubes and detectors for enhanced imaging of all patients, including young

11

children, patients with renal insufficiency, and those who cannot hold their breath. Due to its low-kV imaging technique, Force broadens CT's application for patients with renal insufficiency and offers an acquisition speed of 737 mm/sec, so an entire adult chest, abdomen, and pelvis study can be done in one second with no breath-holds. In cardiac imaging, Force can obtain an entire study within one- quarter of a heart beat at a temporal resolution of 66 msec, which is the speed required to freeze the fastest-moving anatomy, such as the right coronary artery.

9 "Mock" MR PSTNet. The MRI simulator will allow researchers to acclimatize the subjects to the ‘enclosed’ and loud MRI environment before they actually go into a real scanner. This is especially important for studies involving children.

10 fMRI peripherals All the MRI scanners will be fully equipped with the latest state of the art peripherals for functional imaging including LCD goggles, integrated eye-tracking, fiber optic subject response gloves, pneumatic computerized headphones with microphones as well as a full spectrum of physiological recording probes for ECG, GSR, pulse-Ox etc. There is also a specialized MRI compatible computerized olfacto-meter.

11 Neuro Testing room A sound proofed and independent temperature controlled neuro testing room is located near the 3T MRI for physiological testing such as EEG, ERP and other modalities. This room is also equipped with large monitors for paradigm training and testing.

12

B. SMALL ANIMAL FACILITES - CSM-SC1

1 9.4T MR Bruker. This is a high-resolution rodent only MRI scanner allowing for high-resolution in-vivo imaging of mice and smaller specimens. It is 9.4 Tesla 89-mm bore MRI system operating at a proton frequency of 400 MHz (Bruker, Billerica, MA). The 9.4T is equipped with a mouse respiratory and cardiac sensor connected to a monitoring and gating system (SA Instruments, Inc., Stony Brook, NY). Sedation is administered by an Isoflurane/O2 gas mixture delivered through a nose cone and placed in a 30 mm birdcage coil with an animal handling system. Additionally, a temperature controller is available in the bore of the magnet, to maintain the animal in the RF coil at a selected temperature. Recent upgrades (Bruker Paravision 4) have enabled the use of navigator pulses to allow for cardiac and/or respiratory gating without the use of electrodes.

2 7.0T MR Bruker Biospec 70/30. This is a high-resolution MR scanner for small animals. The maximum bore diameter for imaging is 15.4cm. This system is equipped with two gradient choices, a large built-in gradient system with up to 200 mt/m and a slew rate of 640 T/m/s. This gradient in combination with a large circular polarized coil will allow imaging of animals up to 15.4cm in diameter. The system is also equipped with a high performance gradient insert with 440mT/m and slew rate of 3,440 T/m/s for high-resolution imaging. The system has 2 transmit and 4 receive channels. There is a 35mm ID circular polarized coil for in-vivo mouse imaging as well as a 4- channel phased array for mouse brain and a 4 channel phased array for mouse cardiac imaging. There are also 3 other dual tuned 20mm surface coils for 31P, 13C and 19F. The 7T Bruker is equipped with the Autopac system, a fully integrated animal handling, laser guided positioning system. Animal warming holders are available for rats and mice as well as a full spectrum of monitoring peripherals for ECG, triggering and respiratory monitoring etc.

13

3 Biophotonic IVIS-Spectrum.

The IVIS Spectrum in vivo imaging system uses a novel patented optical imaging technology to facilitate non- invasive longitudinal monitoring of disease progression, cell trafficking and gene expression patterns in living animals. The IVIS Spectrum is a versatile and advanced in vivo imaging system. An optimized set of high efficiency filters and spectral un-mixing algorithms lets you take full advantage of bioluminescent and fluorescent reporters across the blue to near infrared wavelength region. It also offers single-view 3D tomography for both fluorescent and bioluminescent reporters that can be analyzed in an anatomical context using our Digital Mouse Atlasor. For advanced fluorescence pre-clinical imaging, the IVIS Spectrum has the capability to use either trans-illumination (from the bottom) or epi-illumination (from the top) to illuminate in vivo fluorescent sources. 3D diffuse fluorescence tomography can be performed to determine source localization and concentration using the combination of structured light and trans illumination fluorescent images. The instrument is equipped with 10 narrow band excitation filters (30nm bandwidth) and 18 narrow band emission filters (20nm bandwidth) that assist in significantly reducing autofluorescence by the spectral scanning of filters and the use of spectral unmixing algorithms. In addition, the spectral unmixing tools allow the researcher to separate signals from multiple fluorescent reporters within the same animal. http://www.perkinelmer.com/Catalog/Product/ID/IVISSPE

4 Micro Ultrasound Vevo 2100 VisualSonics.

This is dedicated Ultrasound system for small animal models (mice to rabbits) of disease. This scanner is capable of all imaging modes found in clinical US scanners such as color Doppler, M-mode, 3D imaging and volume analysis but at much higher spatial resolution. It allows for rapid animal screening of tumor and other models. The higher resolution of this system will also allow for image-guided injection. B-Mode (2D) imaging for anatomical visualization and quantification, with enhanced temporal resolution with frame rates up to 740 fps (in 2D for a 4x4 mm FOV) , and enhanced image uniformity with multiple focal zones. M-Mode is for visualization and

14

quantification of wall motion in cardiovascular research, single line acquisition allows for the very high-temporal (1000 fps) resolution necessary for analysis of LV function. Anatomical M-Mode is for adjustable anatomical orientation in reconstructed M-Mode imaging; software automatically optimizes field of view for maximum frame rate. Pulsed-Wave Doppler Mode (PW) is for quantification of blood flow. Color Doppler Mode is used for detection of blood vessels including flow directional information and mean velocities; as well as for identification of small vessels not visible in B-Mode. Power Doppler Mode is for detection and quantification of blood flow in small vessels not visible in B-Mode; increased frame rates allow for significantly faster data acquisition. Tissue Doppler Mode for quantification of myocardial tissue movement; for example in assessing diastolic dysfunction. Vevo MicroMarker® Nonlinear Contrast Agent Imaging can be used for quantification of relative perfusion & molecular expression of endothelial cell surface markers; enhanced sensitivity to Vevo MicroMarker contrast agents as linear tissue signal is suppressed. 3D-Mode Imaging is for anatomical and vascular visualization, when combined with either B-Mode, Power Doppler Mode or Nonlinear Contrast Imaging; allows for quantification of volume and vascularity within a defined anatomical structure. Digital RF-Mode is for the acquisition and exportation of radio frequency (RF) data in digital format for further analysis; full screen acquisition provides a complete data set for more comprehensive analysis and tissue characterization. ECG and Respiration Gating is used to suppress imaging artifacts due to respiration and cardiac movements. Both are important in cardiac and abdominal imaging for both 2D and 3D data sets. Transducers: * MS-200 12.5 or 21 MHz, Depth from 2mm to 36mm *MS-250 16 or 21 MHz, Depth from 2mm to 30mm *MS-400 24 or 30 MHz, Depth from 2mm to 20mm *MS-550D 32 or 40 MHz, Depth from 1mm to 15mm http://www.visualsonics.com/vevo2100

5 Near IR Frangioni imager. This rodent scanner is designed to visualize cellular probes that fluoresce in the Near IR region which provides much better tissue penetration than traditional Green Fluorescent Proteins.

6 Radiochemistry. This laboratory is equipped radiochemical preparations and calibration.

15

C. Nanomedicine Laboratory - CSM-7th Floor.

This laboratory has 2 modules: the synthetic lab and the analytical/biochemistry/biology lab. We are able to synthesize established imaging reagents for supply and distribution. In the synthetic lab, there are 2 large synthetic chemistry hoods that can accommodate 4 synthetic chemists working simultaneously. Each scientist has individual bench space for work-up and for storing samples, reagents, buffers and the like. The analytical/biochemistry/biology lab has 2 smaller hoods for doing wet chemistry work. Both facilities have been equipped with state-of-the-art instruments to support the work. The synthetic lab is well equipped for investigators to perform small-scale syntheses of organic, inorganic and organometallic compounds for use in a multitude of imaging modalities as well as drug delivery nanoparticles. In addition, we are also capable of labeling peptides and antibodies with commercially available optical dyes, CT, or MR contrast agents.

16

D. Image Analysis and Data Center - CSM- 1st Floor.

One of the main functions of TMII is to provide the infrastructure for access to research imaging. A comprehensive set of Image modalities are supported for both human as well as animal work. Scheduling support for access to the different scanners consist of web-based online calendars as well as live telephone scheduling support. TMII also provides a central hub for image distribution and archival. There are 32TB of online storage where all imaging data is pushed to and distributed from. The capacity will be expanded annually as needed.

1. Image Analysis Core. TMII provides image analysis for cardiovascular, body/oncological and neuro imaging support through the Image Analysis Core. The image analysis for specific projects needs to be discussed directly with the TMII core (contact TMII Director Dr. Zahi Fayad.) This core consists of IT personnel, software engineers, imaging physicists, research assistants and other support personnel. Expert consultation for research projects including protocol design, specialized pulse sequences, special image acquisition hardware (coils), custom made functional MRI stimulus hardware are all supported. Comprehensive project based image analysis support is also provided. Modalities supported include PET, MRI, fMRI, DTI and its variants, resting state fMRI. Image analysis training is also supported for those researchers who want to learn more about image analysis in general. Training range from regular classroom based graduate course taught by TMII faculty to hands on training on the use of specific software packages such as FSL, SPM, Brainvoyager and TMII’s own in house developed software packages in all areas (neuro, body, oncology and cardiovascular). The data center has a dedicated server room which houses a larger Mac Server Cluster with 2 x 16TB of initial online storage with direct connectivity to all the imaging modalities in CSM. In addition, there is also an image analysis room equipped with large viewing display and more than 15 high performance workstations open for the researcher to learn or perform image analysis.

Minerva 2. TMII XNAT. TMII XNAT serves as the central point for research data transfer, archive, and sharing. TMII XNAT is built upon a secure database, supports automated pipelines for processing managed data, and provides tools for exploring the data. Only users authorized by the study investigators can access their data. TMII XNAT is fully HIPAA compliant. The TMII XNAT team XNAT provides support for data migration between various DICOM repositories, HIPAA de-identification, image preprocessing, image quality control, and other customized services. Currently TMII XNAT runs on two mirrored Linux servers with 60TB storage space on each. It can host more than 15,000 image sessions with backups. TMII XNAT user training, documentation, and imaging data management consultations are available by request

17

(https://tmii.mssm.edu/xnat). A yearly service support contract has been established with the XNAT developer group from Radiologics Inc.

3. Image reconstruction tools for PET and fast MR imaging. TMII is equipped with a dedicated workstation for PET images reconstructions, such as the Siemens e7-tools and the open-source package STIR (http://stir.sourceforge.net/). The e7-tools are a collection of Microsoft Windows command line programs that allow the processing and reconstruction of Siemens PET emission data both using iterative and analytical algorithms. The software is capable of generating of other correction factors including attenuation, scatter and normalization. The software also allows for listmode histogramming and rebinning. The software is installed on an external computer to reconstruct PET images away from the scanners. STIR software on the other hand is an open source toolbox that offers the same functionalities as the e7tools, but is not limited to the analysis of Siemens PET emission data. Also available is a dedicated workstation housing PET-SORTEO (Simulation Of Realistic Tridimensional Emitting Objects), a simulation tool that uses Monte Carlo techniques to generate realistic PET data from voxelized descriptions of tracer distributions, in accordance with the scanner geometry and physical characteristics.

4. Soma Server. TMII also hosts a dedicated quad 2.2GHz, 8-core processors server for reconstruction of fast MR images techniques, such as compressed sensing, with 1xK20 GPU (expandable to a second GPU). The specifications of the server are : CPUs, 4 Intel Xeon E5-4620, 2.2 GHz (8-Core, HT, 16 MB Cache); RAM: 256GB (16x16GB DDR3-1600 ECC Registered 2R DIMMs) operating at 1600 MT/s Max; Management: Integrated IPMI 2.0 & KVM with Dedicated LAN; Controller: Dual-Port Intel X540 10GbE plus 8 Ports 6Gb/s SAS LSI 2208 HW RAID, and 6 Ports SATA; Hot-Swap Drive - 1: 3TB Seagate Constellation ES.3 (6Gb/s, 7.2K RPM, 128MB Cache) 3.5-inch SATA; Power supply: Redundant 1400W Power Supply with DSC and PMBus - 80 PLUS Platinum Certified; Rail Kit: Quick-Release Rail Kit for Square Holes, 26.5 - 36.4 inches; OS: Linux Ubuntu 13.10; PCIe 3.0 x16 - 1: NVIDIA Tesla Kepler K20 GPU Compute Board PCIe 2.0 x16. Matlab R2013b (www.mathworks.com) is installed on this server and can be used for heavy duty image reconstruction tasks, by exploiting the servers parallel computing capabilities. The server is available to all TMII researchers through remote login.

More information is also available online@ http://tmii.mssm.edu &

Contact: Zahi A. Fayad, PhD – Director, TMII Christopher J. Cannistraci, MS – Program Manager [email protected] [email protected] (212) 824-8452 (212) 824-8466

18

19 20

Biographies of Invited Speakers

21 22

Zahi A. Fayad, PhD, FAHA, FACC, FISMRM Mount Sinai Endowed Chair in Medical Imaging and Bioengineering Professor of Radiology and Medicine (Cardiology) Director, Translational and Molecular Institute Vice chair for Research, Department of Radiology Icahn School of Medicine at Mount Sinai, New York, NY [email protected]

Biography

Dr. Fayad serves as professor of Radiology and Medicine (Cardiology) at the Mount Sinai School of Medicine. He is the founding Director of the Translational and Molecular Imaging Institute; Vice chair for Research, Department of Radiology at the Icahn School of Medicine at Mount Sinai. Dr. Fayad’s interdisciplinary and discipline bridging research - from engineering to biology and from pre-clinical to clinical investigations - has been dedicated to the detection and prevention of cardiovascular disease with many seminal contributions in the field of multimodality biomedical imaging (MR, CT, PET and PET/MR) and nanomedicine. He has authored more than 300 peer-reviewed publications (h-index of 68 accessed 04/16/2016 on Thomson Reuters Web of Science), 50 book chapters, and over 500 meeting presentations. He is currently the Principal Investigator (PI) of three federal grants/contracts funded by the National Institutes of Health’s National Heart, Lung and Blood Institute and National institute of Biomedical Imaging and Bioengineering with a large award from NHLBI to support the Program of Excellence in Nanotechnology. He is also PI on three new NIH sub-contracts with UCSD, Columbia and the Brigham and Women’s Hospital. In addition, he serves as Principal Investigator of the Imaging Core of the Mount Sinai National Institute of Health (NIH)/Clinical and Translational Science Awards (CTSA). He is a PI of one of the 3 projects in the Strategically Focused Prevention Research Network Center grant funded by the American Heart Association (AHA) to promote cardiovascular health among high-risk New York City children, and their parents, living in Harlem and the Bronx. Moreover, he currently leads four pharmaceutically funded multicenter clinical trials for the evaluation of novel cardiovascular drugs.

He is Associate Editor for the Journal of the American College of Cardiology Imaging (JACC Imaging), Section Editor for Journal of the American College of Cardiology (JACC) and Consulting Editor for Arteriosclerosis Thrombosis and Vascular Biology (ATVB) and past associate Editor of Magnetic Resonance in Medicine (MRM). In 2013, he became a Charter Member, NIH Center of Scientific Review,

23

Clinical Molecular Imaging and Probe Development Study Section. In 2015, he chaired the Scientific Advisory Board of the Institut National de la Santé et de la Recherche Médicale (INSERM) PARCC program at the HEGP in Paris.

Dr. Fayad had his engineering trainings at Bradley University (BS, Electrical Engineering ’89), the Johns Hopkins University (MS, Biomedical Engineering ‘91) and at the University of Pennsylvania (PhD. Bioengineering ’96). From 1996 to 1997 he was junior faculty in the Department of Radiology at the University of Pennsylvania. In 1997 he joined the faculty at Mount Sinai School of Medicine.

Dr. Fayad is the recipient of multiple prestigious awards. In 2007 he was given the John Paul II Medal from Krakow, Poland in recognition for the potential of his work on humankind. As a teacher and mentor, Dr. Fayad has been also extremely successful. He has trained over 100 postdoctoral fellows, clinical fellows and students. His trainees have received major awards, fellowships, and positions in academia and industry. In 2008, he received the Outstanding Teacher Award from the International Society of Magnetic Resonance in Medicine (ISMRM) for his teaching on cardiovascular imaging and molecular imaging. In 2009 he was awarded the title of Honorary Professor in Nanomedicine at Aarhus University in Denmark. Recently, he was one of opening speakers at the 2011 97th Scientific Assembly and Scientific meeting of the Radiological Society of North America (RSNA). In 2012, he was invited to give the Henry I Russek Lecture at the 45th Anniversary of the ACCF New York Cardiovascular Symposium. In 2013, he was elected Fellow of the International Society of Magnetic Resonance In Medicine, Magnetic Resonance Imaging, received a Distinguished Reviewer from Magnetic Resonance in Medicine and was selected as an Academy of Radiology Research, Distinguished Investigator In 2014 he received the Centurion Society award from his alma matter (highest award) Bradley University for his bringing national and international credit to his alma matter. In 2014, he received the Editor’s Recognition Award, from the Journal Radiology. In 2015, he was the Dr. Joseph Dvorkin Memorial Lecturer at the Cardiac Research Day of the Mazankowski Alberta Heart Institute, University of Alberta, Edmonton, Canada. In 2015, he became the Mount Sinai Endowed Professor in Medical Imaging and Bioengineering. The Mount Sinai Professorships were established in 2007 by the Mount Sinai Boards of Trustees to honor the achievements and contributions of some of Icahn School of Medicine’s most outstanding faculty. A total of eight Mount Sinai Professorships have been awarded to date in Alzheimer’s Research, Diabetes and Aging, Gene Medicine, Medical Imaging and Bioengineering, Orthopaedics, Orthopaedic Research, Psychiatric Genomics, and Structural Biology. In 2016, he was the Heart & Stroke/Richard Lewar lecturer at the Center of Excellence in Cardiovascular Research in Toronto.

He is married to Monique P. Fayad, MBA and is the proud father of Chloé (14 year old) and Christophe (9 year old) and after spending seven years in Manhattan now lives in Larchmont, runs in Central Park and participates regularly in New York Road Runners races. He also enjoys regular sailing and stand-up board paddling in Larchmont, New York, Connecticut, Rhode Island, Cape Cod, Martha’s Vineyard, Nantucket, Caribbean Islands and beyond.

24

Dennis S. Charney, MD Anne and Joel Ehrenkranz Dean Icahn School of Medicine at Mount Sinai Executive Vice President for Academic Affairs The Mount Sinai Medical Center

Biography

Dennis S. Charney, MD, is Anne and Joel Ehrenkranz Dean of the Icahn School of Medicine at Mount Sinai and President for Academic Affairs for the Mount Sinai Health System. An internationally acclaimed expert in the neurobiology and treatment of mood and anxiety disorders, Dr. Charney has made fundamental contributions to the understanding of the causes of human anxiety, fear, and depression, and to the discovery of new treatments for mood and anxiety disorders.

Under Dean Charney’s leadership, the Icahn School of Medicine at Mount Sinai has risen to, and maintained, its strength among the top 15 U.S. medical schools in National Institutes of Health (NIH) funding, and currently ranks second in funding per faculty member from all sponsored projects. With an emphasis on innovation and discovery and a track record of strategic recruitments across the biomedical sciences and in genomics, computational biology, and information technology, the School has cultivated a supercharged, Silicon Valley-like atmosphere in the academic setting. As the sole medical school partnering with the seven hospitals of the Mount Sinai Health System, the Icahn School of Medicine at Mount Sinai has one of the most expansive educational, research and clinical footprints in the nation.

Early in his tenure as Dean, Dr. Charney unveiled Mount Sinai's $2.25 billion strategic plan, which laid the foundation for establishing multidisciplinary research institutes as hubs for scientific and clinical collaboration. Within and across the institutes, faculty investigators and physicians work together to push the boundaries of science and medicine in order to address the most pressing biomedical challenges of our time. Dr. Charney is now overseeing the creation of complementary clinical institutes for the entire Mount Sinai Health System. These new institutes are Centers of Excellence for disease-specific areas such as cancer, heart disease, diabetes, HIV, and pulmonary diseases. Together the research and clinical institutes are generating game-changing models in translational research, clinical excellence and standards of care.

Recent affiliations with Rensselaer Polytechnic Institute, Google, IBM and Apple further enhance the landscape

25

for discovery at Icahn School of Medicine at Mount Sinai. These unique relationships have expanded opportunities for cross-fertilization of ideas and programs, and present exciting educational, scientific and clinical possibilities for our students and faculty alike.

Dean Charney's career began in 1981 at Yale University School of Medicine, where, within nine years, he rose from Assistant Professor to tenured Professor of Psychiatry. While at Yale, he chaired the National Institute of Mental Health (NIMH) Board of Scientific Counselors, which advises the institute's director on intramural research programs. In 2000, the NIMH recruited Dr. Charney to lead their Mood and Anxiety Disorder Research Program — one of the largest programs of its kind in the world —and the Experimental Therapeutics and Pathophysiology Branch. That year, Dr. Charney was elected to the National Academies of Medicine.

In 2004, Dr. Charney was recruited to Icahn School of Medicine at Mount Sinai as Dean of Research. In 2007, he was appointed Dean of the School and Executive Vice President for Academic Affairs of the Medical Center. In 2013, Dr. Charney was named President for Academic Affairs for the Health System. He is currently one of the longest-serving Deans of any American medical school.

Dr. Charney’s own robust research program has garnered recognition through virtually every major award in his field. His investigations of the causes and treatment of depression have generated new hypotheses regarding the mechanisms of antidepressant drugs and have resulted in novel therapies, including Lithium and Ketamine for treatment-resistant depression. The work of his research team demonstrating that Ketamine as a rapidly acting antidepressant has been hailed as one of the most exciting developments in antidepressant therapy in more than half a century.

A prolific author, Dr. Charney has written more than 600 publications, including groundbreaking scientific papers, chapters, and books. His many books include: Neurobiology of Mental Illness (Oxford University Press, USA, Fourth Edition, 2013); The Peace of Mind Prescription: An Authoritative Guide to Finding the Most Effective Treatment for Anxiety and Depression (Houghton Mifflin Harcourt, 2004); The Physician’s Guide to Depression and Bipolar Disorders(McGraw-Hill Professional, 2006), Resilience and Mental Health: Challenges Across the Lifespan (Cambridge University Press, 2011).

Dr. Charney is a committed educator and role model who lectures within Mount Sinai, nationally and internationally. He has mentored and taught scores of junior faculty, postdoctoral fellows, medical students and graduate students throughout his career.

26

Burton Drayer, MD, FACR Professor of Radiology Chair of the Department of Radiology Icahn School of Medicine at Mount Sinai

Biography

Burton Paul Drayer, MD is currently the Dr. Charles M. and Marilyn Newman Professor and Chairman of the Department of Radiology (1995-present) at the Icahn School of Medicine at Mount Sinai and the Executive Vice President for Risk at The Mount Sinai Medical Center. Additionally, from 2003 to 2008, Dr. Drayer served as President of The Mount Sinai Hospital. He also serves as CEO of the Mount Sinai Doctors Faculty Practice and Dean for Clinical Affairs at the Icahn School of Medicine at Mount Sinai. He completed his internship and Neurology residency at the University of Vermont and then a Radiology residency and Neuroradiology fellowship at the University of Pittsburgh Health Center. He is Board certified in both Neurology and Radiology and a fellow of both the American College of Radiology and the American Academy of Neurology.

Dr. Drayer served as Associate Professor and Professor of Radiology at Duke University from 1979 to 1986 where he was also Director of Neuroradiology. In 1986, he joined the Barrow Neurological Institute as Director of Magnetic Resonance Imaging and Research. Internationally known for his CT and MRI research on the aging brain and neurodegenerative disorders, brain infarction, multiple sclerosis, and physiological and functional brain imaging, Dr. Drayer has written over 200 publications as well as multiple book chapters. He was the first to describe metrizamide encephalopathy, nonradioactive xenon enhanced CT for measuring rCBF (ASNR Cornelius G. Dyke Award 1977) and the normal and abnormal distribution of brain iron using MRI. He also popularized carotid and intracranial MRA and educated a generation of physicians in the efficient clinical use of brain and spine MRI. He has been on numerous editorial boards and was the editor of Neuroimaging Clinics of North America from 1990 to 2005.

Dr. Drayer was elected President of the ASNR in 1996, was the inaugural Chairman of its Research Foundation, and was awarded the ASNR Gold Medal in 2011. In 2003, Dr. Drayer was elected to the Board of Directors of the RSNA and in 2009 ascended to Chairman of the Board, 2010 President elect, and 2011 RSNA President. He is presently Chairman of the Board of Trustees of the RSNA Research and Education Foundation, is a past- President of the New York Roentgen Ray Society, and has served on numerous national advisory boards for multiple sclerosis, stroke, and Alzheimer’s disease.

27 28

Bruce Fischl, PhD Professor of Radiology Harvard Medical School Director of the Computational Core At the Athinoula A. Martinos Center, MGH Visiting scientist at MIT CSAIL

Biography

Bruce Fischl has a B.A. in Mathetmatics and Computer Science from Wesleyan University, and a Ph.D. in Cognitive and Neural Systems from Boston University where he worked with Dr. Eric Schwartz on instantiating a model of the human retino-cortical visual system on a small mobile robot. Since the end of 1996 he has been at the Athinoula A Martinos Center for Biomedical Imaging which is part of Mass. General Hospital and Harvard Medical School, first as a postdoctoral fellow with Dr. Anders Dale and currently as a full Professor in the Department of Radiology. He is the primary developer of the FreeSurfer suite of neuroimaging analysis tools that has almost 25,000 licenses distributed to clinicians, neuroscientists and engineers. He has published over three hundred papers, 7 of which have been cited more than 1,000 times, and is generally interested in neuroanatomical/functional modeling, improving clinical care by providing enhanced information to neuroradiologists and neurologists, and in developing state-of-the-art imaging for in vivo and ex vivo anatomical modeling with the goal of better understanding the neuroanatomical and neurophysiological basis for human intelligence and how they go awry due to disease processes.

Keynote

“Computational analysis of functional, connectional and architectonicproperties of the human brain with translational applications”

Abstract

Magnetic Resonance Imaging technology has progressed at an astonishing rate in the last decade, with an array of new image contrasts available at higher resolution, higher SNR and/or reduced imaging time. In this talk I will discuss ongoing work at MGH with the goal of automatically extracting information from these images for the purposes of quantifying normal brain structure, function and connectivity as well as detecting departures from normal trajectories. This includes the optimization of

29

the acquisition for postprocessing, the construction and use of surface-based models of the human cerebral cortex, segmentation of an array of non-cortical structures, and the automated modeling of major white matter fascicles from diffusion-weighted MRI. For each of these applications I will show clinical relevance. Finally, I will describe work using ex vivo MRI and optical imaging to directly resolve structures that are far smaller than can be seen in vivo, with resolutions approaching 1 micron, that enables us to probe stereological, laminar and architectonic properties of the human brain with minimal distortion and large fields of view. These properties can then be predicted from macroscopic geometry that can be observed in-vivo, making them available for neuroscientific and clinical research.

30

Julie Price, PhD Professor of Radiology University of Pittsburg

Biography

Dr. Julie Price is Professor of Radiology and Biostatistics at the University of Pittsburgh. Her expertise is in quantitative PET methodology for translational imaging of protein targets, blood flow and glucose metabolism including studies of aging, neurodegeneration, neuropsychiatric disorders and brain injury. Her primary research involves PET imaging of amyloid-beta and (more recently) tau protein deposits in the validation and application of biomarkers of Alzheimer’s disease. She has established expertise in PET pharmacokinetic modeling that includes evaluation of 11C-labeled Pittsburgh Compound-B (PiB) that is one of the most widely used PET amyloid-imaging agents. She is a member of the Alzheimer’s Disease Neuroimaging Initiative (ADNI) PET Core. She is co-author of 140 research articles.

Julie received a B.S. in physics and M.S. in medical physics from the University of Wisconsin. She completed doctoral training in radiation health sciences at Johns Hopkins University and post-doctoral training at the NIH PET/Nuclear Medicine Department. She joined the University of Pittsburgh in 1994. She is Head of PET methodology that includes simultaneous PET/MR neuroimaging. In 2014, she was Visiting Professor of Radiology at the MGH Martinos Center (March-August). She was Chair of the NIH Clinical Neuroscience and Neurodegeneration study section (2013-2015).

Neuroimaging Session

“In vivo PET imaging of protein targets in Alzheimer’s disease”

Abstract

Projections of the future public health burden of Alzheimer’s disease (AD) continue to fuel efforts to better understand AD neurodegeneration in the context of specific neuropathological hallmarks of AD

31

(i.e., amyloid-beta (Aβ) plaques and neurofibrillary tangles composed of hyperphosphorylated tau protein). In vivo imaging of protein targets is possible using PET imaging because of the development of radiolabeled positron-emitting molecules that bind with high affinity and selectivity to protein targets of interest. PET innovations over the past 15 years have enabled in vivo imaging of Aβ plaques and greatly advanced understanding of AD pathophysiological processes in tandem with findings of abnormal structural and functional MRI, FDG PET metabolism, Aβ and tau cerebrospinal fluid concentrations, and cognitive performance. This presentation will provide an overview of the translational research path for one of the most widely used PET Aβ imaging agents, 11C-lableled Pittsburgh compound-B with an emphasis on methodology validation. A review of research advances in the field of PET Aβ imaging will also be presented that include current research and challenges in the field. As it is now possible to also selectively image tau protein deposits with PET, early results and challenges that have emerged in this rapidly expanding area of PET imaging will also be discussed. A key goal continues to be sensitive early detection of AD pathophysiology and identification of those who might benefit most from therapy.

32

Mark Griswold, PhD Professor of Radiology Case Western Reserve University Director of the Interactive Commons (IC)

Biography

Mark Griswold, PhD, is a professor in the Department of Radiology at Case Western Reserve University and University Hospitals in Cleveland, Ohio, with secondary appointments in Biomedical Engineering, Physics, Electrical Engineering and Computer Science. Dr. Griswold is known for his research in developing fast and quantitative MRI methods, including the recently developed Magnetic Resonance Fingerprinting (MRF). Dr. Griswold received his BS in Electrical Engineering from the University of Illinois and his PhD in Physics from the University of Wurzburg, Germany. Prior to joining Case Western Reserve, Dr. Griswold was director of the RF Coil Development Laboratory at Beth Israel Deaconess Medical Center/Harvard Medical School. He is a fellow of the American Institute of Medical and Biological Engineering (2012) and the International Society of Magnetic Resonance in Medicine (2009) where he also serves on the Board of Trustees and is currently the annual meeting program chair.

Cancer & Body Imaging Session

“Rethinking the way we do MRI: Magnetic Resonance Fingerprinting”

Abstract

With very few exceptions, all MR studies since the mid-1960s have used the same basic experimental formulation: a fixed "pulse sequence" generates an MR signal is repeated multiple times to encode spectral, spatial or decay information (or some combination of these quantities). One then applies a mathematical transform such as a Fourier transform (or similar) to obtain the results in "real" quantities, such as frequency or space. This framework of MR works so well that it has remained nearly constant for almost 50 years.

33

In this talk we will discuss a new framework, Magnetic Resonance Fingerprinting (MRF), that we believe has the potential to overcome previous limitations and open up numerous new possibilities for MR. MRF is based on a different concept for MR acquisition, processing and visualization. Instead of using a single "purified" pulse sequence, MRF uses a pseudorandomized pulse sequence which is simultaneously sensitive to multiple parameters during a single acquisition. This provides a rich signal but one that no longer fits into the standard MR processing framework. Following the acquisition of this largely incoherent signal, MRF uses pattern recognition to decode the acquired data. As we demonstrate, MRF is able to recover high quality quantitative results for multiple MR parameters simultaneously from a single acquisition. The pattern recognition used in MRF also provides a high level of suppression of measurement errors stemming from patient motion, measurement noise and undersampling and in certain cases may provide higher sensitivity than traditional MR methods. MRF is also able to directly generate maps specific to individual tissue types, which allows for the characterization of specific tissues when they only occupy a fraction of a pixel, which should allow for earlier disease detection. Finally, MRF should practically simplify the clinical MR workflow, with the potential that the end user could just be presented with a single "scan" button.

34

Anna Moore, PhD Professor of Radiology Harvard Medical School Director, Molecular Imaging Laboratory MGH/MIT/HMS Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital

Biography

Dr. Anna Moore holds a PhD in Bioorganic Chemistry. She joined the Department of Radiology at Massachusetts General Hospital in 1991 as a postdoctoral fellow and is now a Professor of Radiology and the Director of Molecular Imaging Laboratory at the Athinoula A. Martinos Center for Biomedical Imaging at MGH. She made major contributions to developing image-guided nucleic acid-based therapies in cancer. Her group was the first to image siRNA delivery to tumors using dual magnetic resonance and optical imaging. Additional research by her group seeks to develop methods for non- invasive imaging of diabetic pancreas to foster development of more effective therapies to treat this disease. For her contributions to the field of molecular imaging Dr. Moore was elected a Member of the Board of Trustees and the Executive Committee of the World Molecular Imaging Society. She was named a Distinguished Investigator by the Academy of Radiology Research in 2014. Dr. Moore published her research in high impact journal including Nature Medicine, Nature Photonics, Cancer Research, Diabetes, Radiology and others. She is a recipient of multiple grant awards from NIH and other agencies.

Nanomedicine Session

“Image-guided Precision Nanomedicine for Cancer Therapy”

Abstract

Precision medicine is now at the forefront of cancer treatment. The idea of using patient's biological information could improve and tailor treatment to the patient’s individual needs presenting a prime example of personalized medicine. Potential on small non-coding RNAs in that regard is indisputable,

35

considering that one can use this mechanism to silence virtually any gene, with single-nucleotide specificity. Small interfering RNAs (siRNA) and microRNAs have emerged as regulators of post- transcriptional modification of gene expression and are posed to yield extremely promising candidates for cancer therapy. Molecular imaging can provide vital information about the delivery of RNA-based drugs to the tumor site and assist in evaluating the therapeutic outcome. Nanoparticle carriers for these drugs can serve a dual role as delivery vehicles and imaging reporters due to their innate magnetic properties. This presentation will focus on developing nucleic acid-based cancer therapies and their application for treating metastatic disease.

36

Debiao Li, PhD Professor of Radiology Cedars - Sinai Professor of Medicine and Bioengineering University of California, Los Angeles

Biography

Debiao Li, PhD received his PhD in Biomedical Engineering at the University of Virginia in 1992. He was an Assistant Professor of Radiology at Washington University in St. Louis (1993-1998), Associate Professor (1998-2004), Professor (2004-2010) of Radiology and Biomedical Engineering, and Director of Cardiovascular MR Research (2004-2010) at Northwestern University, Chicago. Since 2010, Dr. Li is the Inaugural Director of the Biomedical Imaging Research Institute at Cedars-Sinai Medical Center, Los Angeles. He is also Professor of Medicine and Bioengineering at the University of California, Los Angeles. Dr. Li has conducted cardiovascular MR imaging research for over 20 years, published more than 200 research articles in peer-reviewed journals, and received multi-million dollars of research funding from National Institute of Health. Dr. Li is Past President of the International Society for Magnetic Resonance in Medicine (ISMRM), Past President of the International MR Angiography Working Group, Past President of the Overseas Chinese Society for MR in Medicine, and a former member of Board of Trustees, Society for Cardiovascular Magnetic Resonance. He is also Associate Editor of Magnetic Resonance in Medicine (MRM) and the Journal of Magnetic Resonance Imaging (JMRI). He is a fellow of ISMRM and American Institute for Medical and Biological Engineering. He is widely recognized as a leader in cardiovascular MR research, including coronary artery imaging, atherosclerosis imaging, myocardial blood flow measurement, and non-contrast MR angiography.

Cardiovascular Imaging Session

“MR Coronary Angiography and Vessel Wall Imaging”

Abstract

Coronary artery disease remains the leading cause of deaths in the world. X-ray conventional angiography is the standard diagnosis method. The goal is to detect > 50% luminal stenosis. However,

37

as many as 70% of myocardial infarctions or sudden coronary deaths are caused by plaque rupture of relatively small coronary lesions that are not diagnosed by x-ray angiography. Atherosclerosis is characterized by the build-up of lipid-rich plaques on arterial wall. It affects large and medium sized systemic arteries: aorta, coronary, carotid, and peripheral arteries. It is a systemic disease underlying the majority of cardiovascular disease, including myocardial infarction, stroke, aortic aneurysm, and peripheral vascular disease. Despite considerable technical challenges, MRI methods have been developed to detect and characterize coronary atherosclerosis, including plaque morphology (vessel wall thickening and positive remodeling), plaque composition (lipid core and hemorrhage), and plaque biology/activity (neovascularization and inflammation). In this presentation, these techniques and preliminary studies will be reviewed. In particular, a new 3D T1-weighted whole-heart “hot spot” MR technique will be introduced. It provides interleaved and registered “dark blood” and “bright blood” isotropic resolution images within 10 min. The “dark blood” images highlight hemorrhage and inflammation, respectively before and after contrast administration, and the “bright blood” images provide coronary artery anatomical background. Potential values of coronary plaque imaging include individualized risk stratification to identify asymptomatic subjects at high risk for atherosclerosis progression, who might benefit from drug therapy and lifestyle modifications in early stages of the disease, and patients at high risk of a cardiovascular event in the near future who are in most urgent need of intervention; and serving as surrogate outcomes for clinical trials evaluating therapeutic response.

38

Selected Abstracts for Oral Presentation

39 40

DANTE-EPI for CSF Suppression in Cervical Spinal Cord BOLD fMRI at 7T

Alan C. Seifert1,2, Hadrien A. Dyvorne1,2, Joo-won Kim1,2, Bei Zhang3, and Junqian Xu1,2,4 1Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY USA 2Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY USA 3Department of Radiology, NYU Langone Medical Center, New York, NY USA 4Department of Neuroscience, Icahn School of Medicine at Mount Sinai, New York, NY USA

Introduction Spinal cord blood oxygenation level-dependent (BOLD) fMRI studies have begun to demonstrate convincing activation in the spinal cord gray matter for sensory and pain signals and during resting state. Among many challenges in spinal cord fMRI, physiological noise due to pulsating cerebrospinal fluid (CSF) surrounding the spinal cord is arguably the most detrimental to BOLD signal detection in the spinal cord. Most current spinal cord fMRI studies address the CSF signal contamination issue with image post-processing methods (e.g. regressing out CSF signal, or physiological cardiac recording). In this work, we adapt a CSF signal suppression strategy, delay alternating with nutation for tailored excitation (DANTE), to reduce CSF signal contamination in spinal cord BOLD fMRI at 7T.

Methods & Results The DANTE MRI method consists of a train of low-flip angle RF pulses interleaved with gradient pulses, resulting in suppression of moving spins. In this work, DANTE is combined with an echo-planar imaging (EPI) sequence and applied to the cervical spinal cord at 7T. DANTE with 250 pulses at 15° achieves 95% suppression of CSF signal, while more conservative 150x15° and 150x10° yield 92% and 80% suppression, respectively. These three DANTE conditions, however, result in attenuation of signal in the spinal cord by 42%, 34%, and 26%, respectively. Temporal cross-correlation of resting-state signal in the bilateral spinal cord gray matter improves from r=0.25 (without DANTE) to r=0.80-0.90 (with DANTE). In addition, application of DANTE (250x15°) reduces the temporal correlation of CSF versus gray matter signal significantly (from r=0.54 to r=0.08), suggesting that signal fluctuation related to the pulsating CSF was attenuated and/or removed from the gray matter by DANTE.

Conclusion The present results show that motion-sensitive DANTE preparation strongly suppresses signal from pulsating CSF in the cervical spine at 7T. Because DANTE also causes attenuation of signal in static tissues proportional to Np and α2 [5], these parameters must be selected carefully to balance CSF suppression with preservation of spinal cord signal. The strongest DANTE preparation applied in this work, 250x15°, suppressed CSF most strongly, but at an increased cost to spinal cord signal intensity and tSNR. A more moderate preparation, 150x10°, maintains effective CSF suppression (see Fig. 1), but attenuates spinal cord signal less heavily, and preserves a stronger bi- lateral correlation in gray matter signal than 250x15° (r=0.90 vs. r=0.80, respectively).

Clinical Relevance DANTE prepared EPI with CSF-attenuation/suppression is a promising spinal cord BOLD fMRI acquisition technique at ultra-high field.

41

Figures and tables

Figure 1: Montage of mean absolute signal intensity (grayscale) and temporal SNR (color) for 15 combinations of DANTE pulse train lengths and flip angles and one with no DANTE preparation applied (parameters inset within each frame). In images with strong DANTE preparation (e.g., 250x15°), no signal is visible from CSF, although signal intensity and temporal SNR in the spinal cord are moderately reduced.

Figure 2: Surface plots of the mean absolute signal intensity within the gray matter (mesh edges solid) and CSF (mesh edges dashed) ROIs. As the number of DANTE pulses is increased (i.e., as the DANTE pulse train is lengthened), and as the DANTE RF pulse flip angle is increased, signal from the moving CSF spins is strongly suppressed, while signal from the stationary spinal cord is only moderately attenuated.

Figure 3: Surface plot of the temporal correlation coefficient, r, for signal in ROIs in the left versus right gray matter (note reversal of x and y axis limits vs. Figures 2 and 3). As signal from CSF is more strongly suppressed, signal in the left and right gray matter during resting-state scans becomes more strongly correlated. This suggests the suppression of nuisance signal contributions from the pulsating CSF spins, while the underlying resting-state BOLD signal in the gray matter is preserved.

42

Assessment of tumor heterogeneity in hepatocellular carcinoma using combined DCE-MRI and BOLD measurements

Authors & Affiliations Stefanie Hectors1,2, Mathilde Wagner1,2, Octavia Bane1,2, Cecilia Besa1,2, Nelson Chen1,2, Romain Remark3, M. Isabel Fiel4, Hongfa Zhu4, Meriam Merad3, Bachir Taouli1,2 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States 2 Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States 3 Oncological Science, Icahn School of Medicine at Mount Sinai, New York, NY, United States 4 Department of Pathology, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction Tumor perfusion and oxygenation are two factors that can substantially influence treatment outcome in cancer patients. MRI offers techniques to noninvasively image perfusion and oxygenation, specifically using DCE-MRI and blood oxygen level dependent (BOLD) MRI. The objective of our study was to assess perfusion and oxygenation heterogeneity in hepatocellular carcinoma (HCC) lesions. To that aim, we performed DCE-MRI and BOLD-MRI in HCC patients and assessed the correlation between these techniques both in regions of interest (ROIs) in the entire tumor as well as in tumor segments.

Methods & Results Twenty-eight HCC patients (M/F 23/5, mean age 59y) with 35 HCC lesions (average size 4.6±3.4 cm) underwent abdominal MRI at 1.5 or 3.0T, including BOLD-MRI (axial multiecho gradient echo T2* measurement before and after a 100% oxygen challenge) and DCE-MRI [3D FLASH with 100 dynamics (temporal resolution 2.3 s) during injection of 0.05 mmol/kg Gd-BOPTA]. DCE-MRI parameter (arterial flow Fa, portal flow Fp, total flow Ft, mean transit time MTT, distribution volume DV, arterial fraction ART) maps in the liver, including HCC lesions, were processed by fitting a dual- input single compartment model to the dynamic curves in each pixel, using pre-contrast T1 data acquired in a separate Look-Locker acquisition. R2* measurements pre and post O2 were obtained by monoexponential fitting of the signal data at the different echoes. R2* and DCE-MRI parameters maps were coregistered using landmark-based registration in Matlab. ROIs encompassing the entire HCC lesions were drawn on original DCE-MRI images. Mean, median, kurtosis and skewness values of the DCE-MRI parameters, R2* pre and post O2 in these ROIs were determined. In addition, the ROIs were divided into 1, 4, or 9 segments dependent on lesion size (≤ 3 cm, 3-8 cm and > 8 cm, respectively). Mean and median values of all parameters in each segment were determined. Spearman correlations between the calculated quantitative DCE-MRI and R2* measures were assessed for both the global and segmented ROIs. For 14/35 lesions, the global ROI parameters were also correlated with vascular fractions that were determined from a CD31-stained tumor section using custom threshold-based Matlab scripts.

Representative DCE-MRI and BOLD parameter maps in an HCC tumor are shown in Figure 1, showing substantial heterogeneity. In the global ROIs mean and median DCE-MRI and BOLD parameters were not correlated, while kurtosis and skewness showed some moderate to strong correlations (Table 1). However, in the segmented ROIs mean and median Fa correlated significantly, yet weakly, with both R2* pre and post O2 (Table 2). Mean and median R2* pre O2 in the global ROIs also showed a significant moderate negative correlation with the histology-derived vascular fraction (r=- 0.569, P=0.037 and r=-0.582, P=0.032, respectively).

Conclusion Our results indicate that histogram and segmented ROI analysis of DCE-MRI and BOLD-MRI data of HCC tumors can potentially identify tumor regions with distinct perfusion and oxygenation.

Clinical Relevance

43

MRI-aided spatial identification of tumor perfusion and oxygenation could aid in treatment stratification and monitoring.

Figures and tables

Figure 1 Representative DCE-MRI, R2* and T1 parameter maps of an 8.3 cm HCC lesion in a 54-year-old male patient with hepatitis B virus, showing perfusion and oxygenation heterogeneity. A distinct tumor region of high Fa and Ft and low MTT, DV, R2* pre and post O2 is indicated by white arrows on the Fa and R2* pre O2 maps. This region likely reflects well- perfused and oxygenated tumor tissue. Another area of the tumor, indicated by black arrows, is characterized by low Fa and Ft and high MTT, DV, R2* pre and post O2, likely resembling poorly perfused and hypoxic tumor tissue.

Table 1 Significant correlations between kurtosis and skewness of DCE-MRI and BOLD parameters in global HCC ROIs Kurtosis Skewness Parameters r p Parameters r p

Fa / R2* pre O2 0.436 0.009 Fa / R2* pre O2 0.377 0.026 Fa / R2* post O2 0.409 0.015 Fp / R2* pre O2 0.410 0.015 Ft / R2* pre O2 0.565 <0.001 Ft / R2* pre O2 0.472 0.005 Ft / R2* post O2 0.443 0.008 DV / R2* pre O2 0.351 0.039

Table 2 Significant correlations between mean and median of DCE-MRI and BOLD parameters in segmented HCC ROIs Mean Median Parameters r p Parameters r p

Fa / R2* pre O2 -0.236 0.009 Fa / R2* pre O2 -0.223 0.014 Fa / R2* post -0.201 0.026 Fa / R2* post -0.215 0.017 O2 O2

44

Augmenting drug-carrier compatibility improves tumor nanotherapy efficacy

Y. Zhao1, F. Fay1, S. Hak2, J.M. Perez-Aguilar3, B.L. Sanchez-Gaytan1,Zahi, A.Fayad1, C. Pérez-Medina1, W.J.M. Mulder1 1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY,USA. 2Dep. of Circulation and Medical Imaging, Norwegian Univ. of Science and Technology, Trondheim, Norway. 3Dep. of Physiology and Biophysics, Weill Cornell Medical College of Cornell Univ., New York, NY, USA.

Introduction Drug-nanocarrier association after intravenous administration is essential for efficient drug delivery to the tumor. However, a large number of currently available carriers are self-assembled polymeric nanoparticles whose drug-loading stability is critically affected by the in vivo environment. The drug’s hydrophobicity and its miscibility with the polymeric matrix are known to determine nanoparticle drug loading. However, it remains unclear how these properties contribute to the drug-nanocarrier association in circulation, subsequent tumor delivery efficiency, and resulting therapeutic efficacy. Systematically investigating the effect of drug hydrophobicity and miscibility in vivo is therefore an imperative step towards improving nanoparticle therapeutics.

Methods & Results We built a dual fluorescently-labeled nanoparticle that allowed us to monitor the drug-carrier association using Förster resonance energy transfer (FRET; Fig. a).[1] Through rational derivatization, we were able to fine-tune a model drug’s hydrophobicity and miscibility. We thoroughly studied model drug release rates from the nanoparticle carrier in in vitro systems (Fig. b, c), by computer simulations, and in a mouse tumor model by in vivo FRET imaging (Fig. d, e). The findings were used to define guidelines to augment doxorubicin’s compatibility with nanoparticle carrier using pro-drug derivatization technology.

Conclusion Drugs loaded in self-assembled nanoparticles are - when exposed to serum –susceptible to undesired and premature release. This is due to drug exchange with plasma proteins, including albumin and HDL, and diminishes the nanoparticle’s drug delivery efficiency. We found that augmenting drug-carrier compatibility improves the in vivo stability and eventually drug accumulation in tumors. Based on these findings, we propose a general guideline (Fig. f) for increased doxorubicin delivery efficiency and improved anti-tumor efficacy.

Clinical Relevance Our findings on the relationship between drug-carrier compatibility and drug delivery efficiency prove to be important not only for understanding nanomedicines’ in vivo fate, but also for guiding improvements in nanoparticle drug delivery strategies.

45

Figures and tables

Figure. a, FRET nanoparticles. b, Spectral change upon drug release. c, Dynamic measurement on release in FBS. d, Intravital microscopy images of release in vasculatures. e, NIR imaging of release in tumor. f, A guideline for efficient drug delivery: change drug’s hydrophobicity and miscibility to reduce premature release.

References: 1. Zhao Y. et al, ACS Nano 2013 7 (11), 10362.

46

Lack of improvement in aortic vascular inflammation is associated with an increase in coronary plaque burden in psoriasis

Authors & Affiliations: Lerman JB, Aberra TA, Dey AK, Kabbany MT, Joshi AA, Silverman J, Salahuddin T, Ahlman MA, Playford MP, Chen MY, Bluemke DA, Mehta NN. National Heart, Lung & Blood Institute, NIH, Bethesda, MD.

Introduction: Psoriasis is a chronic inflammatory skin disease that is associated with elevated cardiovascular risk, and increased vascular inflammation (VI) by 18-FDG PET/CT. Previously we have shown that aortic VI correlates with quantified coronary plaque burden in psoriasis. However, the impact of longitudinal changes in VI on the progression of coronary disease is yet unknown.

We hypothesize that a longitudinal improvement in aortic VI may associate with a stabilization of coronary plaque burden in psoriasis. Conversely, we hypothesize that a longitudinal increase in VI correlates with a worsening in coronary plaque burden.

Methods & Results: Consecutively recruited psoriasis patients (N=67) underwent 18-FDG PET/CT, and coronary CT Angiography (CCTA) (320 detector row, Toshiba), at baseline and one-year follow-up. Aortic VI was assessed as target-to-background ratio (TBR). Total (TB), and non-calcified (NCB), coronary plaque burden were quantified using QAngio (Medis, Netherlands). The longitudinal change in coronary plaque burden was analyzed with unadjusted and adjusted regression.

The cohort was middle-aged, at low Framingham Risk, and had mild to moderate psoriasis (Table 1). Patients whose VI worsened (ΔTBR +11%, p <0.001) had a progression in TB (β=0.56, p = 0.03), and NCB (β=0.53, p = 0.04), beyond adjustment for traditional cardiovascular risk factors, waist-hip-ratio, statins, and systemic/biologic psoriasis treatment. Conversely, patients whose VI improved (ΔTBR -11%, P <0.001), had a non-significant change in TB (β=0.06, p = 0.54), and NCB (β=0.07, p = 0.47).

Conclusion: Increasing aortic VI was associated with worsening TB, specifically NCB, at one year. These results suggest that FDG PET/CT measures of vascular inflammation may approximate longitudinal changes in coronary artery disease. However, larger studies are needed to validate these results.

Clinical Relevance: Our study suggests that longitudinal changes in vascular inflammation may approximate the progression of non-calcified coronary plaques. These results are consistent with the known association between VI by FDG PET/CT and prospective cardiovascular events. Additionally, these findings suggest that VI by FDG PET/CT may be found to be a strong surrogate marker of coronary disease advancement.

47

Table 1: Characteristics of the study groups, stratified by improvement vs. non-improvement in aortic VI at 1 year

Parameter Worsening in VI (N = 13) Improvement in VI (N= 54) Baseline 1-year P* Baseline 1-year P* Demographic/Clinical Characteristics Age, years 51.8±12.3 53.1±12.1 - 50.7±12.1 51.8±12.1 - Males, N (%) 8 (61.5%) - - 34 (63.0%) - - Hypertension, N (%) 2 (15.4%) 2 (15.4%) 1 18 (33.3%) 14 (25.9%) 0.13 Type-2 Diabetes, N (%) 2 (15.4%) 2 (15.4%) 1 6 (11.1%) 4 (7.4%) 0.5 Hyperlipidemia, N (%) 7 (53.8%) 7 (53.8%) 1 30 (55.5%) 29 (53.7%) 0.74 Current Smoker, N (%) 1 (7.7%) 1 (7.7%) 1 11 (9.1%) 4 (7.3%) 0.32 Body Mass Index, kg/m2 31.4±6.9 32.2±7.5 0.06 29.6±5.4 29.4±5.1 0.28 Clinical and Lab Values

Systolic Blood Pressure, mm Hg 122.4±12.4 118.1±12.3 0.15 126.5±12.0 117.1±13.6 0.001 Total Cholesterol, mg/dL 164.5±27.2 164.8±34.0 0.48 182.3±39.9 182.6±44.3 0.48 HDL Cholesterol, mg/dL 57.2±14.7 60.15±22.3 0.21 52.7±17.0 57.7±22.8 0.003 HOMA-IR, (Median [IQR]) 2.99 (2.0-3.3) 3.04 (2.2-5.1) 0.02 3.42 (1.8-5.3) 3.7 (2.0-5.5) 0.18 Framingham Risk Score (Median [IQR]) 3 (3-4) 2 (1-6) 0.35 4 (1-7) 2.5(1-7) 0.15 Psoriasis Details

PASI Score (Median [IQR]) 5.6 (4.0-8.9) 4.4 (1.8-5.7) 0.06 5.4 (3.0-12.3) 3.3 (2.2-5.6) 0.002 Systemic/Biologic Therapy, N (%) 9 (69.2%) 11 (84.6%) 0.5 21 (38.9%) 29 (53.7%) 0.005 Lipid Treatment, N (%) 6 (46.2%) 6 (46.2%) 1 18 (33.3%) 19 (35.2%) 0.56 Vascular Inflammation by FDG PET/CT Aortic Target-to-Background Ratio 1.67±0.17 1.86±0.33 0.002 1.82±0.31 1.62±0.19 <0.001

Coronary Plaque Burden

Total Burden (x 100), mm2 1.35±0.53 1.47±0.74 0.04 1.20±0.44 1.21±0.49 0.37 Non-Calcified Burden (x 100), mm2 1.26±0.46 1.43±0.73 0.03 1.15±0.41 1.16±0.49 0.27 Comparisons performed between baseline and 1-year follow up values. Continuous variables analyzed by Student’s t-test, categorical variables analyzed by Pearson’s chi-squared test. P < 0.05 considered statistically significant. VI (Vascular Inflammation), N (Number), HDL (High Density Lipoprotein), HOMA-IR (Homeostasis Model Assessment for Insulin Resistance), IQR- (Interquartile Range), PASI- (Psoriasis Area Severity Index).

48

Abstracts Selected for Poster Presentation Cancer & Body Imaging

49 50

Assessment of interplatform variability of T1 quantification methods used for DCE-MRI in a multicenter QIN phantom study

Authors & Affiliations Bane O1, Hectors S1, Wagner M1, Arlinghaus L2, Aryal M3, Boss M4, Cao Y3, Chenevert T5, Fennessy F6, Huang W7, Hylton N8, Kalpathy-Cramer J9, Keenan K4, Malyarenko D5, Mulkern R6, Newitt D8 , Wilmes L8, Yen Y F9, Yankeelov T10, Taouli B1 1Icahn School of Medicine at Mount Sinai, 2Vanderbilt University, 3University of Michigan Radiation Oncology (2), 4National Institute of Standards and Technology, 5University of Michigan Radiology (1), 6Brigham and Women’s Hospital, 7Oregon Health and Science University, 8University of California San Francisco, 9Massachusetts General Hospital, 10University of Texas-Austin

Introduction T1-weighted dynamic contrast-enhanced MRI (DCE-MRI) is used to quantify perfusion and flow in tumors or other pathologies. The precision of pharmacokinetic parameters estimated from DCE-MRI contrast uptake curves is highly dependent on the conversion of T1-weighted signal to Gd concentration, and thus on the baseline T1 value of the tissue of interest. The objective of this study was to measure interplatform variability in T1 quantification in a multicenter (through the QIN: quantitative imaging network)/NCI study by testing common inversion-recovery spin-echo (IR-SE) and variable flip angle (VFA) protocols using a dedicated T1 phantom.

Methods & Results A T1 phantom, produced by the National Institute of Standards and Technology (NIST), was scanned at 7 different QIN sites on different platforms as listed in Table 1. The phantom consists of 14 spherical vials doped with varying concentration of T1-shortening NiCl2. The T1 mapping protocols were standardized and consisted of an inversion recovery spin echo (IR-SE) and a variable flip angle (VFA) sequence. Acquisition parameters are listed in Table 2. The phantom was scanned twice (test-retest measurements) at each center. Data analysis was performed by a single observer at Mount Sinai and consisted of positioning of circular ROIs centered on each sphere, using OsiriX software. Subsequently, the mean ROI signal was fitted according to the signal equation for each sequence to obtain T1 values, using custom fitting routines written in MATLAB R2015. Test-retest coefficients of variation (CV) and Bland-Altman statistics were computed for the common protocols, for each platform. The standardized VFA protocol was compared with the reference standard IR-SE protocol using Bland- Altman statistics.

T1 measurements in the 14 spheres ranged between 20 and 2000 ms at 3T, as expected, with greater spread in the distribution of T1 values observed between sites with the VFA sequence (Figure 2). The IR-SE protocol had high repeatability at both field strengths, with mean test-retest CV of 0.3% at 1.5T and <7% at 3T (Table 4). The common VFA protocol had poorer repeatability, with test-retest CV <2% at 1.5T, and as high as 18% at 3T (Table 4). The comparison of the common VFA protocol to the IR-SE reference standard protocol across eight 3T magnets showed absolute % difference bias in the range of 2%-36%.

Conclusion Preliminary results show high interplatform variability in T1 values in test-retest scans and between different protocols. Future work will analyze accuracy of each T1 measurement method with respect to gold standard T1 values determined by NMR spectroscopy at NIST. Temperature correction of T1 values will also be performed. The complete dataset will be analyzed with a generalized linear mixed statistical model, to compare accuracy of T1 measurements across field strength, scanner models, and sequences.

Clinical Relevance Standardization of T1 mapping used for DCE-MRI quantification likely improves reliability of estimated pharmacokinetic parameters and thereby potentially enhances the accuracy of e.g. treatment planning and monitoring based on these parameters.

51

Figures and tables

Table 1. Overview of participating sites and magnets Table 2. Standardized acquisition parameters of IR-SE and VFA

Site Magnet IR-SE VFA BWH GE 3T Discovery w750 Orientation Coronal Coronal MGH Siemens 3T Skyra/Tim Trio Flip angle 180 2, 5, 10, 15, 20, 25, 30 MSinai Siemens 1.5 Aera/3T Skyra Echo time (ms) 9 2 OHSU Siemens 3T Tim Trio Repetition time (ms) 5000 12 UMich (1) Philips 3T Ingenia Inversion times (ms) 24, 50, 75, 100, 125, n/a UMich (2) Siemens 3T Skyra 150, 250, 500, 750, Vanderbilt Philips 3T Achieva 1000, 2000, 3000 Field of view (mm2) 200x200 200x200 Number of slices 1 16 Figure 2. Distribution of T1 values in the 14 Slice thickness (mm) 5-6 5-6 phantom spheres at 3T, across 8 scanners. Matrix 256x256 256x256 Echo train length 5-6 n/a Number of averages 1 3 IR-SE T1 values at 3T 3000 Acquisition time (min) 45 13

2500 Table 3. Test-retest repeatability expressed as mean CV (%) at

2000 participating sites for IR-SE and VFA

(ms) 1500 1 T IR-SE VFA 1000 1.5T MSinai Siemens Aera 0.3 1.45 500 3.0T BWH GE Discovery 0.65 4.15

0 MGH Siemens Skyra 0.22 0.93 Sphere MGH Siemens Trio 0.17 10.39 MSinai Siemens Skyra 0.31 18.03 Standard VFA T1 values at 3T 3000 OHSU Siemens Trio 0.25 14.21 UMich1 Philips Ingenia 1.2 15.66 2500 UMich2 Siemens Skyra 6.4 2.31

2000 Vanderbilt Philips Achieva 0.58 1.17

(ms) 1500 1 T

1000

500

0 Sphere

52

Prostate DWI: comparison of a shorter diagonal acquisition to standard 3-scan-trace acquisition

Authors & Affiliations: Idoia Corcuera-Solano, Mathilde Wagner, Stefanie Hectors, Sara Lewis, Nicholas Titelbaum, Bachir Taouli

Introduction Novel DWI sequences used for prostate MRI may allow for reduced acquisition time while maintaining image quality. In diagonal single shot EPI (SS-EPI) DWI (dDWI), the gradients are switched on simultaneously, to maximum amplitude at the highest b-value, which leads to shorter TE compared to standard 3-scan-trace DWI (tDWI). In addition, dDWI is expected to exhibit higher image sharpness, since it uses only one gradient scheme and is therefore less sensitive to differences in eddy currents between gradient schemes. dDWI measures diffusion in one direction (the net gradient direction) as opposed to 3 directions tDWI and is therefore mainly applicable in tissues in which diffusion is considered isotropic. Diagonal DWI has been used previously in a study on the diagnosis of soft tissue tumors (1) and for DWI of the spine (2). We hypothesized that dDWI allows for similar image quality with preserved accurate ADC quantification in a shorter acquisition time compared to tDWI. The aim of our study was to compare dDWI to standard tDWI of the prostate in terms of tumor detection, image quality and quantitative ADC measurement.

Methods & Results 24 consecutive men (mean age 61 y) with suspected prostate cancer underwent 3T MRI (Siemens Skyra) of the prostate including tDWI and dDWI (b-values 50, 1000 and 1600 s/mm2, FOV 250x250 mm2, matrix 114x114, slice thickness 3 mm, 39 slices, TR/TE 8200/69 ms for tDWI and 9300/66 ms for dDWI, averages 1/5/10 in tDWI vs. 2/8/14 for dDWI, average acquisition time 6:21 min for tDWI vs. 4:17 min for dDWI. A higher number of averages was chosen for dDWI to compensate for the loss in SNR in dDWI. Two independent observers evaluated image quality (sharpness, distortion, artifacts and overall quality) on a 5-point scale, ranging from nondiagnostic (1) to excellent (5). ROIs were placed on transitional zone (TZ) and peripheral zone (PZ). Normalized SNR (nSNR) was calculated by dividing mean signal intensity (SI) by the SD of SI in the ROI (3). ADC was measured on both sequences in PZ, TZ and detected tumors. Tumor detection, lesion conspicuity and tumor characterization was also assessed between both DWI sequences. Data was compared between the 2 sequences using paired Wilcoxon signed rank tests and McNemmar test. Coefficients of variations (CV) between ADC measurements were calculated.

Table 1 shows the results from quantitative measurements of SNR and ADC in PZ and TZ. SNR was significantly lower in PZ at b1600 and in TZ at b1000 and b1600 for the dDWI. Mean ADC was significantly higher in TZ with dDWI, while no differences were found for PZ. Reproducibility between sequences was excellent (average CV 3.1±3.0% and 2.7±1.9% for PZ and TZ, respectively). Image quality results are provided (Table 2). Significantly fewer artifacts were observed in the dDWI, while the other image quality scores were similar. There was a non- significant trend towards a greater number of tumor detection with d-DWI for both readers in both b values and ADC maps (Table 3). No significant differences were found in regards of tumor conspicuity and tumor characterization.

Conclusion Diagonal DWI provides substantial reduction in acquisition time (40%) while maintaining adequate tumor detection, image quality and providing equivalent ADC values.

Clinical Relevance d-DWI is an efficient technique decreasing acquisition time while maintaining MRI performance.

Figures and tables Table 1. Quantitative analysis: Quantitative analysis: mean and standard deviation values of SNR for both diffusion sequences in TZ, PZ and tumor and both b values.

53

SNR t-DWI SNR d-DWI p b-1000 TZ 7.09±1.67 5.61±1.53 <0.01 PZ 7.88±3.29 7.17±2.60 0.032

Tumor 7.00±4.26 6.06±3.57 0.49 b-1600 TZ 6.87±1.79 4.86±0.97 <0.01

PZ 7.87±3.17 5.89±2.07 <0.01

Tumor 5.64±2.92 5.62±2.62 0.78

Table 2. Qualitative analysis: mean and standard deviation values of image quality for both diffusion sequences. Reader 1 Reader 2

t-DWI d-DWI p t-DWI d-DWI p b-1000 Image sharpness 4.03±0.63 4.24±0.78 0.07 4.38±0.65 4.29±0.67 0.51 Anatomic distortion 4.50±0.61 4.5±0.75 1.00 4.76±0.61 4.56±0.66 0.04 Artifact 4.44±0.56 4.471±0.66 0.88 4.47±0.51 4.53±0.51 0.48 Sum 12.97±1.36 13.21±1.80 0.20 13.62±1.16 13.38±1.25 0.27 b-1600 Image sharpness 3.70±0.76 3.79±0.81 0.52 3.32±0.94 3.35±0.92 0.88 Anatomic distortion 4.35±0.64 4.29±0.79 0.66 4.35±0.81 4.26±0.83 0.56 Artifact 3.71±0.63 3.79±0.68 0.46 4.20±0.48 4.41±0.49 0.01 Sum 11.76±1.67 11.88±1.87 0.66 11.88±1.70 12.03±1.68 0.66 ADC Overall quality 4.20±0.77 4.176±0.7165 0.84 4.52±0.71 4.32±0.73 0.06

Table 3. Lesion detection assessed with both b values and ADC with t-DWI and d-DWI DWI

b 1000 b 1600 ADC Reader 1 t-DWI 15 (56%) 20 (74%) 22 (81%) d-DWI 15 (56%) 22 (81%) 25 (93%) p* 1.00 0.5 0.25 Reader 2 t-DWI 21 (78%) 23 (85%) 25 (93%) d-DWI 23 (85%) 24 (89%) 26 (96%) p* 0.688 1.00 1.00

Fig 1. t-DWI and d-DWI in a 74 yo patient with prostate cancer. 18 mm lesion in left plPZ in mid gland (PIRADS 5) visible with same conspicuity between t-DWI and d-DWI. Mean ADC of the lesion was 670 for t-DWI vs 653 for d- DWI

54

Engineered Protein Thernanostic Agent for Metastatic Breast Cancer

Joseph A. Frezzo 1, Dung Minh Hoang 2, Youssef Z. Wadghiri 2,3, Jin Kim Montclare1,4 1. Chemical and Biomolecular Engineering, NYU Tandon School of Engineering, Brooklyn, NY, United States, 2. Radiology, NYU School of Medicine, New York, NY, United States, 3. Bernard & Irene Schwartz Center for Biomedical Imaging, New York, NY, United States, 4. Chemistry, NYU, New York, NY, United States

Introduction The field of theranostics merges drug delivery and imaging to promote more effective therapies1. This is especially important in the delivery of cancer therapeutics such as doxorubicin (DOX) where clinicians must strike a balance between administering an effective dose while limiting off-target cardiotoxicity risks2. In this work, a fluorinated protein is investigated for theranostic use as a chemotherapeutic carrier and 19F MRI agent. The fluorinated protein, CE2-RGD-TFL, is comprised of two functional domains: 1) a coiled-coil domain (C), flanked by two integrin targeting domains, capable of encapsulating small hydrophobic drugs and; 2) two elastin-like peptide domains (E) that impart concentration-dependent thermoresponsiveness. In this study, CE2- RGD-TFL shows promise as a thermally triggered T2-weight MRI contrast agent with DOX delivery properties.

Methods & Results Trifluoroleucine (TFL) incorporation was achieved via recombinant expression in leucine auxotrophic E. Coli. Proteins were subjected to UV-Vis spectroscopy, circular dichroism (CD), dynamic light scattering (DLS), inversion recovery and Car-Purcell-Meiboom-Gill (CPMG) 19F NMR at 11.7-T. Phantoms were 19 1 imaged using a 7-T Bruker micro-MRI system tuned to F or H radiofrequencies. CE2-RGD retains structure upon fluorination and also coacervates in the physiological range for hyperthermic treatment (39-42oC). Inversion recovery experiments show little change in R1 values, regardless of temperature or concentration. CPMG experiments reveal a remarkable dependence in R2 values as a function of concentrations and temperatures. Based on r2/r1 as a function of concentration, there is support for using the protein as a T2-nano-thermometer. 1 MRI Phantoms containing CE2-RGD-TFL and water both show positive signals for H nuclei while only the CE2- 19 RGD-TFL showed a positive signal for F nuclei. Finally, CE2-RGD-TFL exhibits 49.1% loading of DOX while CE2- RGD possesses 17.8% loading.

Conclusion Incorporation of TFL in CE2-RGD yields a drug carrier nanoparticle that can undergo temperature dependent structural changes measured by an increase in R2 values which leads us to believe that CE2-RGD-TFL can be used as a R2-dependent thermosensor. Furthermore, a 19F MRI phantom was used to confirm the imaging applicability of CE2-RGD-TFL. Finally, fluorination imparts both greater thermoresponsiveness and greater loading of doxorubicin which provides further value to CE2-RGD-TFL as a theranostic agent.

Clinical Relevance Because of the damaging off-target effects of doxorubicin, there is a necessity for a drug delivery vehicle that can do the following: a) actively target tumor types; b) triggers release of a drug at the tumor site; and c) possesses the ability to be imaged for confirmation of vehicle localization at or near the tumor. Our preliminary biophysical results show promising value of a drug delivery vehicle that could potentially limit off-target effects of doxorubicin and maximize doxorubicin delivery to a tumor.

References 1. Tan, M. et al. Theranostics 2011, 1, 83-101. 2. Rahman, et al. Int. J. of Nanomed. 2007, 2;4: 567-583

55

Figures and tables

Characterization of Fluorinated protein CE2-RGD-TFL A) DLS Temperature trend showing concentration dependent coacervation, B) Inversion recovery 19F NMR yields relatively stable R2 values across temperatures and concentrations while R1 demonstrate a very weak variation. C) CPMG 19F NMR reveals increasing R2 values with temperature and concentration, D) DOX encapsulation efficiency increases upon fluorination, E) MRI phantoms elicit positive 19F signals for fluorinated protein.

56

Diffusion Tensor Imaging of Human Achilles Tendon by Stimulated Echo RESOLVE (ste-RESOLVE)

Authors & Affiliations Xiang He1, Kenneth Wengler2, Alex C Sacher3, Marco Antonio Oriundo Verastegui1, Alyssa Simeone4, Kevin Baker1, Mingqian 1 1 1 Huang , Elaine Gould , and Mark Schweitzer 1Department of Radiology, Stony brook University School of Medicine, Stony Brook, NY, United States, 2Department of Biomedical Engineering, Stony brook University School of Medicine, Stony Brook, NY, United States, 3SUNY Binghamton University, Binghamton, NY, United States, 4New York Medical College, Valhalla, NY, United States Introduction Clinically, Achilles tendon (AT) rapture accounts for 40% to 60% of all operative tendon repairs, with 75% due to sports-related activities. Diffusion tensor imaging (DTI) can be potentially sensitive to the injury-induced changes on the tendons micro-structure (1,2). However, tendons are known to have short T2/T2* relaxation time constant (~15-25 ms for the slow decay component of AT in healthy subjects)(3), so conventional spin-echo based DTI with TE ~80 ms would lead to poor tendon MR signal and difficult diffusion quantification (4,5).

Methods & Results In this study, a novel approach of combining stimulated-echo based DTI (6) with readout-segmented multi-shot EPI (7) (ste-RESOLVE) had been developed and evaluated. The sequence was based from RESOLVE DTI sequence by replacing the bipolar diffusion preparation block with a stimulated echo diffusion prep block. Nulling of eddy current was accomplished by placing a compensation gradient at the end of spin evolving period (TM), while the diffusion encoded spins that were temporary stored at the longitudinal axis would not be affected. Five healthy volunteers were recruited for this IRB approved study. All experiments were performed on 3T Siemens Skyra and Prisma scanners using a flexible 4-channel coil. To boost AT MR signal, subjects were instructed to lay side with the tendon positioned ~55 degrees with respect to magnet B0 field (magic angle). Fig. 1 shows the DWI images acquired by the proposed and the conventional RESOLVE sequence. Higher SNR and better image quality for Achilles tendon can be appreciated in ste-RESOLVE images. Meanwhile, the AT signal in conventional DWI images were too low for robust DTI analysis. For this subject, the mean AT ADC estimated from ste-RESOLVE is 1.1±0.1 mm2/s, mean FA is 0.30±0.1. DTI Studio was used to visualize tendon DTI tensor and generate fiber tractography. Fig. 2 presents the result from one subject. The spatial distribution and orientation of DTI fiber tract follows the parallel collagen fiber bundles running along the major axis of the AT. The proximal portion of the AT fiber tractography demonstrated the muscle-tendon junction. Conclusion We have developed and evaluated a novel stimulated echo-based DTI technique, ste-RESOLVE, capable of achieving high SNR by significant reducing on TE. Robust estimation on ADC, FA and fiber tractography images have been demonstrated. After further optimization and validation, ste-RESOLVE can be used to evaluate early tendon injuries and monitor the healing process of tendon graft. Clinical Relevance Conventional MR images exhibit poor contrast and specificity in delineating low-grade tendon injuries. The proposed ste_RESOLVE techinque will enable, for the first time, a robust investigation the microscopic tissue integrity of Achilles tendon on clinical MR scanners.

57

Figures and tables

Fig 1: DWI of Achilles tendon using the proposed ste-RESOLVE (a) and conventional RESOLVE (b) sequences. Intensities for DWI images acquired with b=800 s/mm2 were doubled.

Fig 2. DTI tractography of Achilles tendon acquired with ste-RESOLVE approach with b value of 400 s/mm2 and 12 diffusion directions. AT DTI fiber track runs along the collagen fiber bundle orientation.

58

Intravoxel incoherent motion diffusion-weighted imaging of hepatocellular carcinoma: is there a correlation with flow and perfusion metrics obtained with dynamic contrast-enhanced MRI?

Authors & Affiliations Stefanie Hectors1,2, Mathilde Wagner1,2, Cecilia Besa1,2, Hadrien Dyvorne1,2, Octavia Bane1,2, M. Isabel Fiel3, Hongfa Zhu3, Bachir Taouli1,2 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States 2 Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States 3 Department of Pathology, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction Intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) allows for simultaneous assessment of tissue diffusion and pseudodiffusion due to capillary blood flow. IVIM-DWI may serve as a surrogate for contrast-enhanced MRI measurements and thereby allow for assessment of tissue perfusion without contrast injection. The aim of our study was to quantify IVIM-DWI and DCE-MRI parameters in hepatocellular carcinoma (HCC) lesions and liver parenchyma and to assess the correlation between these quantitative techniques.

Methods & Results Twenty-five patients with HCC (M/F 23/2, mean age 58y) underwent abdominal MRI at 1.5 or 3.0T, including IVIM-DWI (respiratory-triggered single-shot SE-EPI, bipolar diffusion gradient, b-values 0, 15, 30, 45, 60, 75, 90, 105, 120, 135, 150, 175, 200, 400, 600, 800 s/mm2, TR one respiratory cycle, TE 78-81 ms, FOV 340-450 x 220-305 mm2, reconstruction matrix 320 x 100-256, slice thickness 7 mm, 20-30 slices covering the entire liver) and DCE-MRI (3D FLASH, TR 2.7 ms, TE 0.97-1.07 ms, FA 9.5-11.5°, reconstruction matrix 384x264-312, FOV 370-401x254-325 mm2, 40 slices, slice thickness 4-5 mm, mean temporal resolution 2.3 s, 100 dynamics, contrast agent 0.05 mmol/kg Gd-BOPTA). Regions-of-interest (ROIs) were placed on the IVIM-DWI and DCE-MRI images in the entire HCC lesion and in the liver parenchyma and were matched as closely as possible between the acquisitions. Additional ROIs were drawn in the portal vein and aorta in the DCE-MRI images. IVIM- DWI parameters (pseudodiffusion coefficient D*, diffusion coefficient D and perfusion fraction PF) in the ROIs were quantified by a Bayesian fitting algorithm. DCE-MRI parameters (arterial flow Fa, portal flow Fp, total flow Ft, mean transit time MTT, distribution volume DV and arterial fraction ART) were determined by fitting a dual-input single compartment model to the dynamic data in HCC and liver, using pre-contrast T1 data acquired in a separate Look-Locker acquisition. Differences in IVIM-DWI and DCE parameters between liver and HCC tissue were tested for significance using a Mann- Whitney U test. Spearman correlation coefficients between all IVIM-DWI and DCE-MRI parameters were determined, both for liver and HCC.

IVIM-DWI and DCE-MRI data of 34 HCC lesions (mean size 4.9±3.5 cm) and 25 livers were analyzed. D, D*, PF and Fp were all significantly lower in HCC vs. liver, while Fa and ART were significantly higher in HCC (Table 1). Significant moderate to strong negative correlations were observed between ART and D*, ART and PF, ART and PFxD*, Fa and PF and between Fa and PFxD* in the liver (Figure 1 a-e). There were no significant correlations between IVIM-DWI and DCE-MRI in HCC lesions [highest P-value of 0.086 found for correlation between Fa and PF (Figure 1 f)].

Conclusion IVIM-DWI and DCE-MRI provide non-redundant information in HCC, while they correlate in liver parenchyma.

Clinical Relevance The absence of correlation between IVIM-DWI and DCE-MRI in HCC indicates that the two techniques provide complementary information on tumor vasculature structure, which could be useful for e.g. treatment planning.

59

Figures and tables

Table 1 Mean±SD IVIM-DWI and DCE-MRI parameters in the liver (n=25) and HCC (n=34).

Liver HCC p IVIM-DWI D* 95.9 ± 48.8 63.1 ± 42.5 0.007 D 1.49 ± 0.27 1.35 ± 0.27 0.035 PF 0.23 ± 0.06 0.18 ± 0.08 <0.001 ADC 1.55 ± 0.25 1.52 ± 0.32 0.421

DCE-MRI Fa 56.0 ± 33.0 190.9 ± 189.1 <0.001 Fp 145.8 ± 128.2 69.2 ± 160.6 <0.001 Ft 201.8 ± 126.7 260.1 ± 292.8 0.789 MTT 24.2 ± 9.3 30.6 ± 24.5 0.765 DV 54.0 ± 21.9 49.2 ± 30.4 0.165 ART 33.6 ± 22.8 83.7 ± 22.3 <0.001

Figure 1 Correlation plots between ART and D* (a), ART and PF (b), ART and PFxD* (c), Fa and PF (d) and Fa and PFxD* (e) in the liver and between Fa and PF in HCC (f).

60

MR elastography and DCE-MRI of the liver and spleen for non-invasive prediction of portal pressure

Authors & Affiliations Stefanie Hectors1,2, Mathilde Wagner1,2, Octavia Bane1,2, Thomas D. Schiano3, Aaron Fischman2, Bachir Taouli1,2 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States 2 Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States 3 Division of Liver Disease, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction Portal hypertension (PH) is a common complication of liver cirrhosis. Definitive diagnosis of PH is based on hepatic venous pressure gradient (HVPG) measurement (1-3), an indirect surrogate for portal pressure, which is invasive and not widely available. The goal of this study was to assess whether DCE-MRI and MR elastography (MRE) of liver and spleen provide quantitative biomarkers for the prediction of PH.

Methods & Results Twenty-six patients (M/F 11/15, mean age 50y) who underwent HVPG measurement were included in this prospective IRB- approved study. MRI examination (1.5T and/or 3.0T) was performed within 3 months of HVPG and consisted of MRE of liver and spleen (n=25) and/or DCE-MRI (n=20). MRE was acquired using a 2D GRE sequence using a mechanical motion and motion encoding gradients frequency of 60 Hz. DCE-MRI was acquired using a 3D FLASH sequence with a temporal resolution of 2.3-2.7 s (contrast agent 0.05 mmol/kg Gd-BOPTA). Liver (LS) and spleen (SS) stiffness were determined from stiffness maps. DCE-MRI data were analyzed using model-free parameters (time-to-peak TTP, peak concentration Cpeak, area-under-the-curve at 60 s AUC60 and upslope) and pharmacokinetic modeling [dual-input single compartment model for liver (parameters: arterial flow Fa, portal flow Fp, arterial fraction ART, distribution volume DV, mean transit time MTT), trans Tofts model for spleen (parameters: transfer constant K , extravascular extracellular space ve, rate constant kep)]. Differences in MRI parameters between patients with and without PH were tested for significance with a Mann Whitney U test. MRI parameters were correlated with HVPG using Spearman correlation analysis. ROC for prediction of HVPG ≥5 (PH) and ≥10 mmHg (significant PH) was performed for individual and combinations of parameters.

Mean HVPG was 8.0 ± 7.5 mmHg. There were 14 patients with PH, and 11 with clinically significant PH. LS, SS, liver TTP, spleen TTP and liver upslope were significantly different between patients with and without clinically significant PH, while only LS was significantly different between patients with and without PH (Table 1). There were significant positive correlations between HVPG and liver TTP, liver MTT and LS, while liver upslope was negatively correlated with HVPG (Fig. 1). ROC analysis provided significant AUCs for HVPG ≥5mmHg (LS and SS; Fig. 2 a) and HVPG ≥10mmHg (LS, SS, liver TTP, liver upslope, spleen TTP, spleen upslope and liver MTT; Fig. 2 b). Sensitivity-specificity of LS were 64%-91% and 71%-89% for the detection of PH and significant PH respectively, while the combination of LS and spleen TTP yielded the highest sensitivity-specificity for both the detection of PH and significant PH (92%-86% and 100%-92%, respectively).

Conclusion The liver and spleen perfusion and stiffness metrics can be combined into a multiparametric analysis to maximize diagnostic performance for the prediction of portal pressure.

Clinical Relevance Non-invasive multiparametric MRI combining DCE-MRI and MRE may eventually replace invasive portal hypertension measurements and could be applied repetitively to diagnose and monitor portal hypertension.

Presentation category, please mark your preference [ x]Cancer /Body [ ]Cardiac [ ]Nano [ ]Neuro

Figures and tables

61

Table 1 Mean±SD MRE and DCE-MRI parameter values in patients with and without (significant) PH. Only parameters for which a significant difference between patients groups with HVPG < 5 mmHg and HVPG ≥ 5 mmHg (PH) or between groups with HVPG < 10 mmHg and HVPG ≥ 10 mmHg (significant PH) was found are shown. HVPG < 5 HVPG ≥ 5 p HVPG < 10 HVPG ≥ 10 p LS (kPa) 3.32 ± 2.12 5.62 ± 3.32 0.017 3.61 ± 1.90 7.19 ± 3.97 0.012 SS (kPa) 6.21 ± 1.69 8.70 ± 3.27 0.053 6.82 ± 2.54 9.53 ± 3.14 0.048

Liver TTP (s) 40.4 ± 15.1 55.0 ± 19.4 0.097 41.4 ± 12.6 67.2 ± 19.7 0.015 Spleen TTP (s) 20.2 ± 15.4 35.3 ± 25.9 0.232 20.8 ± 14.9 49.1 ± 27.8 0.019 Liver upslope (mM/s) 0.013 ± 0.006 0.009 ± 0.005 0.133 0.013 ± 0.005 0.006 ± 0.003 0.012

Figure 1 Correlation between HVPG and liver LS (a), liver TTP (b), MTT (c) and liver upslope (d).

Figure 2 ROC curves for detection of PH (a) and clinically significant PH (b).

62

Longitudinal MEMRI Characterization of a Novel Mouse Medulloblastoma Model

Authors & Affiliations Harikrishna Rallapalli1, 2, Eugenia Volkova1, I-Li Tan3, Alexandre Wojcinski3, Alexandra L. Joyner3, Daniel H. Turnbull1, 2 1Skirball Institute, New York University School of Medicine, New York, NY, United States, 2Developmental Biology, Sloan Kettering Institute, New York, NY, United State, 3Department of Radiology, New York University School of Medicine, New York, NY, United States

Introduction In vivo imaging modalities provide powerful tools for the noninvasive longitudinal characterization of preclinical cancer models. Medulloblastoma (MB) is the most common malignant brain tumor in children, and the subject of intense research, much of which involves mouse models. Manganese-enhanced magnetic resonance imaging (MEMRI) produces unparalleled images of the cerebellum, the site of most MBs [1,2]. For this reason, longitudinal MEMRI of preclinical medulloblastoma models enables analysis of the region of origin, monitoring of tumor progression, and treatment response evaluation. In this study, we present the initial MEMRI characterization of a novel mouse medulloblastoma model with an activating mutation in the Smo gene, which exhibit different growth characteristics than those observed in previous studies of Ptch1 knockout mice [1].

Methods & Results SmoM2 mice were engineered by crossing Atoh1-CreER [3] male mice with homozygous R26-floxedSTOP-SmoM2 females [4]. The SmoM2 mutation was induced by subcutaneous injection of low dose (1µg/g) Tamoxifen (TMX) at postnatal day P2. Biweekly imaging sessions using 7-Tesla MRI (Bruker) began at postnatal day P21. MnCl2 (50-60 mg/kg) was injected intraperitoneally 24 hours before imaging. Scan protocol: 1 min low-resolution pilot, 20 min 150µm resolution T1-weighted GE sequence (TE/TR = 4/30 ms; FA = 20°; FOV = 19.2 mm × 19.2 mm × 12 mm; Matrix = 128 × 128 × 80). Images were analyzed in 3-space using Amira and Fiji. Morphological characterization was corroborated with histology as shown in Fig1.

Longitudinal MEMRI results are summarized in Fig2. Based on our preliminary results, all SmoM2 mice had pre-neoplastic lesions, while approximately half developed into full tumor morphology (n=21). Of the mice with tumors, approximately 72% developed bilateral tumors and the remaining developed tumors in either the right or left hemisphere. Approximately 50% of animals with bilateral tumors exhibited regression in one lateral tumor and progression in the other, or progression in both tumors (n=8). General disease progression is as follows: at approximately postnatal week W3, small lesions are apparent in the majority of interlobule spaces including the mid vermis; at ~W7, regions of proliferative lesion thickening are apparent and smaller lesions regress; at ~W13 significant tumor encroachment into the forebrain as well as expansion of the third and fourth ventricles are apparent. Tumors were observed to originate in the posterior hemispheres, shift and compress the normal appearing cerebellum as they progress, and finally encroach into the forebrain. Estimated tumor volume doubling time is approximately 4.5 days at early timepoints (

Conclusion / Clinical Relevance

In addition to qualitative understanding of tumor progression, we have manually segmented and quantified tumor volume at these key timepoints in an effort to produce a unified growth model. Current efforts in automated segmentation and hierarchical clustering-based classification of tumors will guide upcoming preclinical trials of anti-cancer therapeutics.

63

Figures and tables

Figure 1. MEMRI results show clinically relevant structural abnormalities. Key features (arrows) of tumor (T) growth are apparent in both MEMRI and histology in sagittal section. A) expansion of the fourth ventricle B) abnormal lesion proliferation in the anterior right hemisphere C) compression of lobule in right hemisphere D) significant tumor encroachment rostrally into the forebrain, up to the dentate gyrus.

Figure 2. Longitudinal disease state monitoring using MEMRI. 2D and 3D visualization of disease state at key timepoints is made possible by MEMRI.

64

Diagnostic value of computed high b values for prostate cancer detection

Mathilde Wagner1 MD PhD, Idoia Corcuera-Solano1,2 MD, Sara Lewis2 MD, Martin Kang1, Stefanie Hectors1 PhD and Bachir Taouli1,2 MD 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 2 Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction Diffusion weighted imaging (DWI) plays a central role in prostate cancer detection. Recently, high b values have been added to the acquisition protocol in the hopes of improving detection. In parallel, the computed diffusion images have been developed, mathematically derived from the signal intensity decay curve with increasing b-values. The aims of this study were to compare image quality and diagnostic performance of calculated and acquired b 1600, and to assess the added value of calculated b2000 for prostate cancer detection.

Methods & Results 33 men with 54 pathologically proven prostate cancer and 7 control patients were retrospectively included (mean age 61 yo). All patients underwent 3T MRI (Siemens Skyra) of the prostate including trace DWI (b-values 50, 1000 and 1600 s/mm2, FOV 250x250 mm2, matrix 114x114, slice thickness 3 mm, 39 slices, TR/TE 8200/69). Using Olea Sphere software, a computed b1600 image and an ADC map (ADC2b) were constructed using b 50 and b 1000 acquired data. A computed b2000 image and an ADC map (ADC3b) were constructed using b 50, b 1000 and b 1600 acquired data. Three sets of images were assessed by two independent observers: set 1 (acquired DWI b1600/ADC3b), set 2 (set 1 + computed b2000) and set 3 (computed b1600 and ADC2b). T2WI, T1WI, DWI b50/b1000 and DCE-MRI were made available for all sets. The observers evaluated image quality of DW images (acquired b1600, computed b1600, computed b2000) including sharpness, distortion, noise, artifacts and overall quality and ADC map quality using a 5-point scale, ranging from nondiagnostic (1) to excellent (5). For each identified lesion, the observers had to assess lesion conspicuity, PI-RADS v2 DWI score and global PI-RADS v2 score. The reference standard for tumor diagnosis was the analysis by a different radiologist using MRI and pathological data. Lesions were considered to be a match when occurring within the same position of the prostate in terms of all evaluated location attributes. Image quality, lesion conspicuity and tumor characterization were compared using paired Wilcoxon signed rank tests between datasets.

There was no difference in total image quality scores between acquired and computed b1600. The quality of computed b 2000 was significantly higher than computed b1600 for one observer, as well as the quality of ADC3b compared to ADC2b. Observer 1 detected 34/50 (63%) and observer 2 detected 36/50 (67%) of tumors. There was no difference in lesion classification according to PIRADS v2 global score between the 3 sets. However, for DWI PI-RADS score, set 2 had a significant larger number of PI-RADS 4 lesions compared to set 1 (P=0.014 and 0.008 for observer 1 and 2, respectively) and compared to set 3 for observer 2 (P=0.034) (Table 1). The lesion conspicuity score is presented in Figure 1. The lesion conspicuity of computed b2000 was significantly higher than that of computed b1600 and of acquired b1600 (P<0.019). The lesion conspicuity of computed b1600 was significantly higher than the one of acquired b1600 for observer 2 (P=0.009) and not different for observer 1 (P=0.126).

Conclusion Computed b1600 had similar quality and diagnosis performance than acquired b1600. Computed b2000 showed better lesion conspicuity and had similar diagnosis performance than DWI using b1600 but there was a trend of increased performance with the upgrade of some lesions from PIRADS 3 to PIRADS 4.

Clinical Relevance Computed b1600 could be an alternative to acquired b1600 to decrease the acquisition time. Computed b2000 could be added to the standard protocol to increase diagnosis performance.

65

Figures and tables Table 1: DWI PIRADS scores from two observers in 33 patients with 54 tumors.

Reader 1 Reader 2 SET 1 SET 2 SET 3 SET 1 SET 2 SET 3 PIRADS 1 (2%) 1 (2%) 1 (2%) 0 (0%) 0 (0%) 0 (0%) 2 PIRADS 16 (30%) 10 (18%) 12 (22%) 10 (19%) 3 (6%) 9 (17%) 3 PIRADS 14 (26%) 20 (37%) 18 (33%) 18 (33%) 25 (46%) 19 (35%) 4 PIRADS 3 (6%) 3 (6%) 3 (6%) 8 (15%) 8 (15%) 8 (15%) 5 Not seen 20 (37%) 20 (37%) 20 (37%) 18 (33%) 18 (33%) 18 (33%) P value* 0.014 (set 1 vs. set 2) 0.008 (set 1 vs. set 2)

0.102 (set 1 vs. set 3) 0.705(set 1 vs. set 3) 0.480 (set 2 vs. set 3) 0.034 (set 2 vs. set 3) Set 1: acquired b1600 and ADC 3b; set 2: set 1 + computed b2000; set 3: computed b1600 + ADC 2b. All sets included T1W, T2W and DCE images. * Wilcoxon signed rank tests.

Figure 1: Lesion conspicuity, based on a 5-point scale (1 not seen, 5 best contrast compared to normal PZ) from 2 observers using acquired and calculated b values. Reader 1 Reader 2

40 40

30 30 1 2 20 20 3 4 10 10 5 Number of cases Number of cases

0 0 Acq b1000 Acq b1600 Calc b1600 Calc b2000 Acq b1000 Acq b1600 Calc b1600 Calc b2000

66

Technical failure of liver MR elastography exams: Single center experience

Authors & Affiliations Mathilde Wagner1 MD PhD, Idoia Corcuera-Solano1,2 MD, Grace Lo2 MD, Steven Esses2 MD, Joseph Liao2 MD, Nelson Chen2, Ginu Abraham2 RT, Cecilia Besa1,2 MD, Maggie Fung3, James S. Babb4, Richard L. Ehman MD5 and Bachir Taouli1,2 MD 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States 2 Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States 3 GE Healthcare, MR Applications & Workflow, New York, NY, United States 4 Department of Radiology, New York University, New York, NY, United States 5 Department of Radiology, Mayo Clinic, Rochester, Minnesota, United States

Introduction Magnetic resonance elastography (MRE) is an emergent non-invasive method to stage liver fibrosis. The aim of the study was to assess the determinants of technical failure of liver MRE in a large series of patients.

Methods & Results 781 MRE examinations in 691 consecutive patients (mean age 58 y, 63% male) performed in a single center between 6/2013 and 8/2014 were retrospectively evaluated. MRE was acquired either on a 3.0T (n=443) or 1.5T system (n=338), using a Gradient-Recalled-Echo (GRE) MRE sequence. Image analysis was performed by two observers. Technical failure was defined as no pixel with confidence index higher than 95% and/or no apparent shear waves imaged. Logistic regression analysis was performed to assess the link between MRE technical failure and potential predictive factors of failure. The technical failure rate for MRE exams at 1.5T was 4%, while it was higher, 15%, at 3.0T. On univariate analysis, BMI, liver iron deposition, massive ascites, use of 3.0T, presence of cirrhosis, alcoholic liver disease were all significantly associated with MRE failure (P<0.004); while on multivariable analysis, only BMI, liver iron deposition, massive ascites and use of 3.0T were significantly associated with MRE failure (P<0.004). Presence of steatosis, metallic artifacts and subcutaneous fat thickness had no significant impact on failure rate (P=0.113- 0.510).

Conclusion MR elastography with a GRE-based sequence at 1.5T had a low technical failure rate. Use of a GRE-based MRE sequence at 3.0T resulted in a substantially significant higher technical failure rate. Massive ascites, iron deposition, and high BMI were also independent factors associated with liver MRE failure.

Clinical Relevance These results provide motivation for the use of alternative sequences at 3.0T, such as spin-echo echo planar techniques, which are less affected by susceptibility effects.

67

Figures and tables Figure: 53 yo male patient with HCV cirrhosis. Liver MRE was successful at 1.5T with large liver coverage by pixels with a confidence index higher than 95% on the confidence map (A) and good wave propagation (B), while repeat MRE (162 days after) failed at 3.0T with no pixel with a confidence index higher than 95% on the confidence map (C) and no wave propagation (disorganized waves) on the wave image (D). Patient had iron deposition on both exams (T2* = 7.5 ms at 3.0T and 17.2 ms at 1.5T) and had overweight (BMI = 26/27.6).

68

Abstracts Selected for Poster Presentation Cardiovascular Imaging

69 70

Usefulness of combined FDG-PET/MR to diagnose active cardiac sarcoidosis.

Authors & Affiliations Ronan Abgral1-2, Marc R. Dweck1-3, Philip M. Robson1, Maria G. Trivieri1, Nicolas Karakatsanis1, Venkatesh Mani1, Javier Sanz4, Johanna Contreras4, Valentin Fuster4, Maria Padilla5, Jason C. Kovacic4, Zahi A. Fayad1.

1- Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States. 2- Department of Nuclear Medicine, European University of Brittany, EA3878 GETBO, IFR 148, Brest, France. 3- British Heart Foundation, University Centre for Cardiovascular Science, University of Edinburgh, Edinburgh EH16 4SB, United Kingdom. 4- Cardiovascular Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States. 5- Division of Pulmonary, Critical Care and Sleep Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, United States.

Introduction Sarcoidosis is a granulomatous disease of unknown aetiology that most commonly affects lung and mediastinal lymph nodes. Heart involvement is probably under-diagnosed due to frequent absence of clinical symptoms but poses an increased risk of sudden death. A major current obstacle is the inability to easily diagnose active cardiac sarcoidosis (CS) with a non-invasive method. Several imaging techniques are already used to assess CS including myocardial inflammatory activity using 18F-fluorodesoxyglucose (FDG) Positron Emission Tomography (PET) and the pattern of injury using Magnetic Resonance (MR) with late Gadolinium enhancement (LGE). Recent advances now allow combination of these 2 techniques to benefit from the best of these modalities. Our aim was to assess the usefulness of FDG-PET/MR in the diagnosis of CS.

Methods & Results Patients with clinical suspicion of cardiac sarcoidosis were referred in our department for PET/MR imaging. PET data were reconstructed using a Dixon MR attenuation correction map. Myocardial PET uptake was quantified using 3 different methods: SUVmax (maximal standardized uptake value); TBRmax (target to blood pool ratio); and TNRmax (target to negative LGE myocardium ratio). A final diagnosis of active cardiac sarcoidosis (CS+) or no active (CS-) was defined by a consensus of clinical experts with access to all clinical, imaging and biopsy data. Mean SUVmax, TBRmax and TNRmax in CS+ and CS- patients were compared using a Student t-test. 23 patients (12M/11F; 54.6 ± 9.9 yo) were prospectively included from August 2015 to March 2016. One of them did not perform the exam due to claustrophobia. All others underwent PET/MR 69.0 ± 8.9 min after injection of 4.8 ± 0.2 MBq/Kg of FDG. Six patients were considered as CS+ and 16 as CS-. Mean SUVmax were respectively 3.6 ± 1.7 and 6.3 ± 5.6 in CS+ and in CS- patients (p = 0.275). Mean TBRmax were respectively 1.7 ± 0.6 and 2.9 ± 2.4 in CS+ and in CS- patients (p = 0.258). Mean TNRmax were respectively 1.6 ± 0.5 and 1.1 ± 0.1 in CS+ and in CS- patients (p = 0.0001). A clear threshold of TNRmax = 1.25 accurately differentiated all patients as being CS+ or CS-. Two previously unknown cases of sarcoid involvement in bone and in liver were also identified. In the CS- group, combined FDG-PET/MR identified an alternative cause for the cardiac symptoms in 4 (25%) patients (1 arrhythmogenic right ventricular cardiomyopathy, 1 incidental chronic infarction, 1 aortic valve fibroelastoma, 1 anomalous right coronary origin with malignant course).

Conclusion The results of our prospective study show the usefulness of combined FDG-PET/MR in the diagnosis of active CS using TNRmax measurements with clinical impact in patient management. This series also confirms the clinical

71

ability of FDG-PET/MR imaging to evaluate extra-cardiac involvement of sarcoidosis and in assessing alternative myocardial pathologies.

Clinical Relevance Active cardiac sarcoidosis is difficult to diagnose. Combined FDG PET/MR imaging holds major clinical potential in improving the diagnosis and guiding the treatment of patients.

Figures and tables

72

Optimization of 3 dimensional (3D), high resolution T2 weighted SPACE for carotid vessel wall imaging on a 7T whole-body clinical scanner

Authors & Affiliations Gilles Boeykens 1,2,3 , Bram Coolen3, Bei Zhang1, 2, Philip Robson1, 2, Venkatesh Mani1, 2, Art Nederveen3, Zahi Fayad1, 2, Claudia Calcagno1, 2

1Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 2Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 3Academisch Medisch Centrum, Amsterdam, Netherlands

Introduction: Accurate morphological measurements and classification of carotid plaques require imaging with high spatial resolution, and may therefore benefit from increased signal-to-noise ratio (SNR) intrinsically available on ultra-high field (7T) systems1-3. Several studies have already investigated carotid vessel wall imaging at 7T and compared it with state-of-the-art 3T protocols1-3. These initial investigations have focused on 2D multi- slice imaging. Better than this approach, 3D vessel wall imaging allows characterizing extensive vascular territories while minimizing partial volume artifacts in plaque-prone regions, such as the carotid bulb and bifurcation4-5. Here, we present our initial experience of 3D carotid vessel wall imaging on a whole body 7T clinical magnet using a custom made carotid coil. Methods MRI acquisition: 5 volunteers were imaged on a 7T whole body clinical scanner (Siemens Magnetom) using a dedicated, custom designed, 8 channel carotid coil (Figure 1), consisting of 2 transmit elements and 4 receive elements on each side of the neck. Black blood vessel wall imaging was performed using 3D weighted SPACE (Sampling Perfection with Application optimized Contrast using different flip angle Evolutions) with 5 different acquisition settings as detailed in Table 1. MRI analysis: Inner and outer vessel wall contours were delineated on axial slices using Osirix software An additional noise region outside of the neck was also drawn. SNR in the vessel wall was calculated as the signal intensity in the wall divided by the standard deviation of the noise region in the same axial slice. Vessel wall to lumen contrast-to-noise ratio (CNR) was calculated as the difference between vessel wall and lumen SNR. Statistical analysis: A Wilcoxon paired test was used to compare SNR and CNR between 0.8 mm3 and 0.6 mm3 coronal T2W SPACE with isotropic resolution, while a Friedman paired test was used to evaluate difference between the 3 axial acquisitions (0.6, 0.5 and 0.4 mm2 with anisotropic voxels). p values less than 0.05 were considered significant. Results: Images from all subjects and all acquisitions were of sufficient image quality for further analysis. Figure 2 shows curved multi-planar reconstructions from the 0.8 and 0.6 mm3 coronal acquisitions with isotropic voxels, with zoomed in view of the carotid vessel wall at the bottom (orange star indicates the vessel lumen). Figure 3 shows representative images from axial acquisitions (A, 0.6 mm2; B, 0.5 mm2; C, 0.4 mm2), and depicts examples of vessel wall (orange circle), lumen (green circle) and noise (yellow circle) tracings. Average vessel wall SNR (0.8 mm3: 20.7 +/- 11.18; 0.6 mm3: 12.6 +/- 5.8) and vessel wall/lumen CNR (0.8 mm3: 14.6 +/- 8.1; 0.6 mm3: 9.0 +/- 4.5) measurements were significantly different between the two coronal acquisitions with isotropic voxels (Figure 4). Vessel wall SNR was significantly different between the 3 axial acquisitions (0.6 mm2: 36.9 +/- 21.0; 0.5 mm2: 25.3 +/- 11.5, 0.4 mm2: 21.6 +/- 10.9) while CNR measurements indicated no significant difference (0.6 mm2: 22.4 +/- 13.1; 0.5 mm2: 16.7 +/- 7.1, 0.4 mm2: 15.1 +/- 7.7) (Figure 4). In all cases SNR and CNR were comparable to values reported in the literature for similar acquisitions performed on 3T magnets, while achieving equivalent or higher spatial resolution.

73

Conclusion : In conclusion, we demonstrate feasibility of 3D imaging of the carotid vessel wall at ultra-high field using 3D T2W SPACE. SNR and CNR measurements indicate good image quality and good vessel wall/lumen delineation even at high spatial resolution. These results warrant investigation of these techniques in patients with carotid artery disease, and the development of these protocols for multi-contrast and quantitative imaging of the carotid vessel wall at 7T. Clinical Relevance: Even though 3T magnets deliver good image quality of atherosclerotic plaque in the carotids, the higher magnetic field of the 7T scanner might be able to give more insight on the earlier stages of plaque composition.

Figures and tables

Figure 4: SNR (A) and CNR (B) measurements across left and right carotid for all subjects are represented as box plots. Boxes encompass 1st and 3rd quartiles. Lines within each box indicate median SNR and CNR. Error bars indicate the minimum and maximum values. 0.8ISO, coronal 0.8 mm ; 0.6ISO, coronal 0.6 mm ; Figure 1: Custom made, 8-channel 0.6Tra, axial 0.6 mm ; 0.5Tra, axial carotid coil on the table of whole- 0.5 mm ; 0.4Tra, axial 0.4 mm . body 7T scanner. Orange parenthesis and star: significant differences (p<0.05). Gray parenthesis: no significant difference. Figure 3: Representative images from the axial acquisitions (A, 0.6 mm ; B, 0.5 mm ; C, 0.4 mm ), and examples of vessel wall (orange circle), lumen (green circle) and noise (yellow circle) tracings.

References 1. Seven-tesla magnetic resonance imaging of atherosclerotic plaque in the significantly stenosed carotid artery: a feasibility study. de Rotte AA1, Koning W, Truijman MT, den Hartog AG, Bovens SM, Vink A, Sepehrkhouy S, Zwanenburg JJ, Klomp DW, Pasterkamp G, Moll FL, Luijten PR, Hendrikse J, de Borst GJ. Invest Radiol. 2014 Nov;49(11):749-57 2. MRI of the carotid artery at 7 Tesla: quantitative comparison with 3 Tesla. Koning W1, de Rotte AA, Bluemink JJ, van der Velden Figure 2: Curved multi-planar TA, Luijten PR, Klomp DW, Zwanenburg JJ. J Magn Reson Imaging. 2015 Mar;41(3):773-80. reconstructions (MPR) of 3. A transmit/receive radiofrequency array for imaging the carotid arteries at 7 Tesla: coil design and first in vivo results. Kraff O1, Bitz AK, Breyer T, Kruszona S, Maderwald S, Brote I, Gizewski ER, Ladd ME, Quick HH. Invest Radiol. 2011 Apr;46(4):246-54. coronal 0.8 (A) and 0.6 (B) mm 4. Carotid arterial wall MRI at 3T using 3D variable-flip-angle turbo spin-echo (TSE) with flow-sensitive dephasing (FSD). Fan Z1, (upper panel). The lower panel Zhang Z, Chung YC, Weale P, Zuehlsdorff S, Carr J, Li D. J Magn Reson Imaging. 2010 Mar;31(3):645-54 shows zoomed-in, axial 5. Carotid plaque assessment using fast 3D isotropic resolution black-blood MRI. Balu N, Yarnykh VL, Chu B, Wang J, Hatsukami reformatted images of the right T, Yuan C. Magn Reson Med. 2011 Mar;65(3):627-37 common carotid artery (slice level indicated by yellow dashed line). The yellow star indicates the vessel lumen.

74

The diversity of coronary artery and myocardial infarction in mice

Jiqiu Chen, Nadjib Hammoudi, Ludovic Benard, Delaine K. Ceholski, Shihong Zhang, Djamel Lebeche and Roger J. Hajjar

Background: The aim of the present study was to explore the factors that affect the data consistence and potentially lead to low reproducibility in the model of heart failure induced by myocardial infarction (MI) in mice.

Method and results: MI was induced by left coronary artery (LCA) ligation. 61% of LCA were visible and 39% of LCA were invisible in 168 mice under surgical microscope. Echocardiography and in vivo hemodynamics were used to detect the cardiac function after MI. Coronary artery was casted with resin. The morphologic diversity of LCA was visualized with fluorescent imaging. LCA diversity was associated variation of MI size and the LV function by echocardiography and hemodynamics in vivo.

Conclusion: The morphologic diversity and invisibility of LCA is one of major basis of variation in MI size

and location, leading to data deviation which could be one of the primary causes of low reliability and

repeatability in heart failure research in mice.

Clinical Relevance

75

Figure 1. The diversity of left coronary artery (LCA) and myocardial infarction (MI) in mice. A. LCA is not very clear in vivo under surgical microscope. B. LCA ligation and myocardial infarction. C. LCA was casted with resin. D. Fluorescent imaging of LCA and E. Capillaries in ex vivo. F. MI by regular photo. G. MI by echocardiography. H. Hemodynamics of MI. I. Data shows the association of diversity of LCA morphology and variation of cardiac function.

76

MRI Derived Myocardial Strain Analysis for Clinical Evaluation: A Lean Methodology Applied in a Case Study of SERCA2a Gene Therapy Post Myocardial Infarction

Authors & Affiliations Fargnoli AS, Katz MG, Hajjar RJ, Bridges CR Icahn School of Medicine at Mount Sinai, Cardiology

Introduction Cardiac gene therapy trials could greatly benefit by demonstrating regional contractility improvement. Specifically, it is paramount to establish efficacy in diseased myocardium, which is subject to further decline and or recurrent infarction. In trials where MRI is utilized however, resultant parameters are global and typically yield inconclusive results. Advanced cardiac MRI methods to characterize regional function such as tissue tagging or velocity encoding are gold standard research tools. However, from a practical standpoint, these methods are inefficient since they double acquisition time with complex analysis methods. Therefore, these invaluable tools are seldom implemented in gene therapeutic evaluations. Here, we present a more efficient approach that derives accurate myocardial strains from clinically relevant CINE MRI sets in a robust translational model.

Methods & Results A translational gene therapy evaluation in an ovine infarction model (n=12) was executed via surgical ligation (MI) of arterial branches. At 4 weeks post MI, a single treatment group (n=6) received 1 x 1014 viral copies encoding the SERCA2a gene therapeutic with the Molecular Cardiac Surgery Delivery (MCARD) method. MCARD isolates the myocardium in situ via cardiopulmonary bypass and recirculation in a cardiac perfusion circuit of therapeutic for a 20-minute interval. This established system completely separates the heart from the systemic circulation, thus offering a superior high cardiac specific expression profile while limiting off target effects. The remaining (n=6) were designated as controls with no clinical intervention. All subjects had MRI evaluation prior to infarct and terminal 6-month endpoint. Resultant exam CINE, 2CH, 3CH and 4CH were loaded into the SEGMENT Cardiac MRI strain software (Lund, Sweden) to generate regional strain. Myocardium was sectored into healthy remote (Anterior), borderzone (Posterior) and infarcted (Posterolateral) zones as shown in the short (Fig1A) and long (Fig1B) axes. Note, this was in the mid left ventricular zones only where infarct was detected in all cases. Strain was measured from end diastole to end systole in the 3 directions of circumferential, radial and longitudinal axes.

Previously published work demonstrated MCARD mediated delivery of SERCA2a resulted in significant improvement in global function via EF%, LV dimensions, and energetics status. In terms of regional function, all zones in both control and MCARD-SERCA2a treatment suffered decline from Pre MI baseline at 6 months in the radial and circumferential planes. In the longitudinal plane however, the SERCA2a gene treatment group demonstrated significantly higher strain in the diseased borderzone [-14±4] vs. Controls [-6±2], preserving the greatest respective baseline value [-25±4], p<0.01 (Fig1C). The SERCA2a group also trended higher in the infarcted zone [-12±4] vs. Control [-7±3].

77

Conclusion The evaluation of routinely collected CINE, 2CH, 3CH and 4CH sets yielded high quality myocardial strain in the circumferential, radial and longitudinal planes. These results strongly support that robust SERCA2a overexpression drives contractility in diseased myocardium. Significant myocardial infarction in the posterolateral wall induces irreversible changes in radial and circumferential planes. The contractility improvement in the longitudinal plane represents a key reversal in maladaptive remodeling.

Clinical Relevance Myocardial strain evaluation is challenging to implement clinically due to problems encountered with extended cardiac MR acquisition time and challenging quantitative approaches. This approach demonstrates a reproducible methodology to obtain invaluable regional data from routine acquisition sets of CINE, 2-3-4 CH views. This reproducible lean method can be readily implemented in post MI patients in clinical trials.

Figure 1. Quantitative Regional Strain Mapping: A. Short Axis B. Long Axis C. Quantitative Results

78

Increased vascular inflammation in patients with chronic moderate-to-severe atopic dermatitis

Authors & Affiliations Emma Guttman-Yassky1, Patrick M. Brunner2, John Nia1, Peter Hashim1, James G. Krueger2, Philip M. Robson3, Marc R. Dweck3, Venkatesh Mani3, Mark Lebwohl1, Zahi A. Fayad3.

1 Department of Dermatology, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2 Laboratory for Investigative Dermatology, The Rockefeller University, New York, NY, USA 3 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA

Introduction Atopic dermatitis (AD) is the most common chronic inflammatory skin disease, affecting more than 30 million people in the US. Often starting during childhood and adolescence, more than 80% of AD patients have persistent symptoms of chronic itch and skin dryness in adulthood. AD often presents with a wide range of allergic and non-allergic comorbid disorders, including asthma, allergic rhinoconjunctivitis, food allergies, recurrent ear infections, visual impairment, and impaired sleep. Epidemiological studies have recently revealed that patients suffering from AD also have increased risks for cardiovascular disease, but the level of vascular inflammation in these patients has never been investigated.

Methods & Results

In a cohort of moderate-to-severe AD patients with no history of cardiovascular comorbidities (mean age 49, age range 31-66), we investigated inflammation of carotid arteries and the thoracic aorta using FDG-PET/MR. After fasting for at least 6 hours, PET acquisition commenced 60 minutes after injection of 10 mCi FDG and lasted for 30 minutes at the carotid bed and 40 minutes at the thoracic bed positions. FDG and MR images were fused to identify vascular FDG signal and exclude focal signal from adjacent lymph nodes. When quantifying the FDG signal in a 2 cm segment around the carotid bifurcation in left and right arteries and in a 1 cm section of the ascending and descending aorta at the level of the aortic root, the mean target-to-background ratio was increased in all vessels investigated. PET signal was in the same order of magnitude as with chronic cocaine users, serving as a positive control for a patient population known to be at high risk for vascular inflammation, and exhibited a target-to-background ratio (TBR) greater than 1.6, a previously determined threshold for elevated vascular inflammation1.

Conclusion These data show for the first time that chronic AD patients suffer from profound vascular inflammation.

Clinical Relevance The presence of overt vascular inflammation with atopic dermatitis will likely influence future treatment strategies for this patient population.

79

References 1. Mani V, et al. Predictors of change in carotid atherosclerotic plaque inflammation and burden as measured by 18-FDG-PET and MRI, respectively, in the dal-PLAQUE study. The international journal of cardiovascular imaging. 2014;30(3):571-582.

Figures and tables

Figure 1: FDG signal fused with maximum intensity projection of MRI in carotid arteries in a patient with moderate-to-severe atopic dermatitis.

Figure 2: Target-to-background ratios (most diseased segment, MDS) for 5 patients with moderate-to-severe atopic dermatitis (AD), in comparison with a patient population of cocaine use disorder (CUD). LCC: Left carotid artery; RCC: Right carotid artery.

80

Magnetic Resonance Imaging and Positron Emission Tomography to Examine the Cardio-and Cerebrovascular Effects of Welding Fume Exposure

Authors & Affiliations Megan Horton,1 Denise Gaughan,1 Aletheia Donahue,1 Lynn Onyebeke,1 Demetrios Papazaharias,1 Victoria Wang,2 Cheuk Tang,2 Jason Kovacic,3 Mani Venkatesh,2 Robert Wright,1 Roberto Lucchini1

1Department of Preventive Medicine, Icahn School of Medicine at Mount Sinai 2Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai 3Department of Cardiology, Icahn School of Medicine at Mount Sinai

Introduction The process of welding generates nano-size metal particulate containing metals such as manganese (Mn) that can enter the body via the respiratory or olfactory tracts leading to adverse health outcomes. Welders experience increased prevalence of myocardial infarction and other cardiovascular events and decreased neuropsychological function consistent with cerebrovascular injury. Inhalation of Mn particulate from welding may be a risk factor for vascular disease by increasing systemic and local inflammation, defined by accumulation of macrophages. Magnetic resonance imaging (MRI), including Positron Emission Tomography-Magnetic Resonance (PET/MRI), of the carotid artery enables reliable identification of plaque composition. The objective of this study is to use PET/MRI to determine whether exposure to welding fume particulate among steamfitters is associated with cardiovascular and cerebrovascular function.

Methods & Results We obtained PET/MRI images of carotid plaque inflammation, carotid plaque burden, white matter connectivity (using diffusion tensor imaging (DTI)) from 8 New York City Ironworkers. We measured Mn concentrations in brain using quantitative MRI and in toenails using iodine and Mn deposition. We measured Mn concentrations in toenails inductively coupled plasma mass spectrometry (ICP-MS). We used generalized linear models to examine associations between Mn exposure, carotid plaque formation, carotid plaque burden and white matter connectivity.

Conclusion Among these 8 subjects, years of welding and carotid inflammation were inversely associated with functional brain network connectivity (i.e., decreased fractional ansitropy). Additionally, years of welding, lack of personal protective equipment use, and carotid inflammation were positively associated with increased white matter lesions. All of the aforementioned results were statistically significant at p<0.05 prior to adjustment for multiple comparisons. A non-significant trend was observed between basal ganglia Mn deposition and years of welding.

Clinical Relevance Our data suggest an association between increased Mn exposure through inhalation of welding fume particulate and adverse cardiovascular and cerebrovascular outcomes. In 2016, we will enroll an additional 10 subjects to explore these findings in a larger sample size.

81

82

Cardiovascular Inflammation Reduction Trial (CIRT) - Inflammation Imaging Study

V. Mani1, C. Ma1, R. Pyzik1, J. DeLeon6, J. Gaztanaga6, T. Kerwin4, A. Pradhan2, R. Rosenson1, M. Farkouh5, A. Tawakol3, Z.A. Fayad1 1TMII, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2Brigham & Women’s Hospital, Boston, MA, USA 3Massachusetts General Hospital, Boston, MA, USA 4NewYork-Presbyterian/Queens, Flushing, NY, USA 5University Health Network, Toronto, Ontario, CA 6Winthrop University Hospital, Mineola, NY, USA

Introduction Inflammation is a prominent feature of atherosclerotic plaques at high risk for causing cardiovascular disease (CVD) events. The CIRT Inflammation Imaging Study is an ancillary study of the NHLBI funded (Ridker 5U01HL101422) parent CIRT trial that examines whether a reduction in vascular inflammation with an intervention that does not improve other causal pathways in the atherosclerotic process can effectuate a reduction in CVD events. CIRT investigates if an anti-inflammatory agent commonly used in rheumatoid arthritis (low dose methotrexate (LDM)) can reduce CVD morbidity and mortality among patients with a prior myocardial infarction. It is crucial to incorporate a measure of vascular inflammation imaging for confirmation of the primary mechanism of action underlying the CIRT trial. 18-fluoro-deoxy-glucose positron emission tomography/computed tomography (18- FDG-PET/CT) has been established as a reproducible measure of vascular inflammation in both animal and human studies and we propose to use this well validated approach to directly visualize vascular inflammation.

Methods & Results The CIRT Imaging substudy is a double-blinded, randomized, placebo-controlled, parallel-group, multicenter 18-FDG-PET/CT study in patients treated with LDM (15-20 mg/wk). The overall study design is shown in Figure 1. It leverages the existing infrastructure and population recruited as part of the main CIRT trial to recruit a subset of 216 subjects for 18-FDG-PET imaging across 3 metropolitan areas (New York, Boston, and Toronto). Each subject is imaged twice: baseline imaging prior to open-label run-in with LDM and follow-up imaging 8 months after randomization into the main CIRT trial, using identical protocols and equipment in 3 respective imaging centers. At each imaging visit, subjects have their blood drawn for biomarker assessments and blood glucose measurements, and are injected with 10 millicurie of FDG for the FDG-PET/CT scan. Imaging is performed across 3-4 bed positions, covering the neck and chest and CT attenuation correction is performed using tube voltage of 120 kVp, and current of approximately 40 mAs. PET data are acquired in 3-D mode over ~10 minutes per bed position and stored on 256x256 matrixes. Attenuation-corrected images are reconstructed using OSEM algorithm yielding ~4 mm effective resolution. Image analysis is performed after careful assessment of image quality. Using the attenuation CT image that was generated during PET-CT imaging, we identify the vessels of interest (right and left carotid arteries as well as the ascending aorta) and draw around the ascending aorta regions of interest (ROI) to provide a maximum standardized uptake values (SUV) for each ROI. This is repeated along the length of the vessel (every ~3.5 mm along the long axis of the vessel) to provide a stack of ROIs that compose the whole vessel (Figure 3). Then, the background corrected maximum SUVs are averaged to provide a whole vessel target-to background ratio (TBR).

Currently, 7 subjects have been recruited into the study, but only 4 were eligible for CIRT Imaging. 3 subjects have completed baseline imaging and 1 subject is pending imaging. The imaging data acquired so far have been robust and of high quality as initially assessed by the core lab (MSSM). Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number R01HL128058.

Conclusion Recruitment has started and is expected to end by September 2018.

Clinical Relevance This substudy has the potential to provide valuable information with regard to the inflammatory status of the vessel walls of the individuals participating in this trial. This may provide mechanistic insights into the role of LDM in reducing cardiovascular inflammation and thereby provide additional data on future CVD risk.

83

Figures and tables

Figure 1:

Figure 2: Sample images of aorta obtained using 18-FDG PET/CT and corresponding imaging endpoints determined

84

Improvement In Cholesterol Efflux Capacity Is Associated With Improvement In Vascular Inflammation By FDG PET/CT In Psoriasis

Authors & Affiliations: Mehta NN, Joshi AA, Lerman JB, Aberra TA, Dey AK, Kabbany MT, Silverman J, Teague HL, Ng Q, Ahlman MA, Playford MP. National Heart, Lung, and Blood Institute, NIH, Bethesda MD

Introduction: Psoriasis (PSO), a chronic inflammatory skin disease associated with increased cardiovascular (CV) risk, provides an ideal human clinical model to study inflammatory atherogenesis. While PSO is associated with increased vascular inflammation (VI), and impaired cholesterol efflux capacity (CEC), the longitudinal impact of a change in CEC on VI is poorly defined. We hypothesized that an improvement in CEC may lead to an improvement in VI by FDG PET/CT.

Methods & Results: Consecutively recruited PSO patients (N=90) underwent 18-FDG PET/CT scans, and cardiometabolic profiling at baseline and 1-year follow-up. Target-to-background ratio (TBR) was assessed using a dedicated PET/CT workstation (Excellence Brilliant Workstation). CEC was measured using previously published, standardized methods. The cohort was stratified based on improvement in CEC. The longitudinal change in TBR and CEC was analyzed using unadjusted and adjusted multivariable regression. The cohort had low CV risk by Framingham risk score (Median 3, IQR 1-6) and mild-to-moderate PSO. Though TBR improved in both stratified groups, it achieved significance only in the group that improved their CEC (TBR mean±S.D. baseline 1.80±0.32 vs. 1-year follow-up 1.68±0.25; p=0.004). The reduction in TBR significantly associated with improvement in CEC (β=−0.22, p = 0.001) beyond traditional CV risk factors, BMI, statins and PSO skin disease therapy. Upon likelihood ratio testing, improvement in CEC provided the maximum value in the determination of TBR reduction (χ2=12.75, p=0.0004), even beyond improvement in PSO severity and improvement in insulin resistance.

Conclusion: Improvement in CEC associated with reduction in aortic VI assessed as TBR by FDG PET/CT. Our study suggests that an improvement in reverse cholesterol transport may reduce VI, an indicator of early subclinical atherosclerosis. Larger studies are needed to confirm these findings.

Clinical Relevance: Our study suggests the importance of the function of HDL cholesterol beyond its concentration in the blood. Based on published literature this ~7% reduction in TBR would lead to a 20% decrease in future clinical CV events, implicating a potentially significant role of reverse cholesterol transport in CV disease prevention.

85 86

Optimal Selection of Thresholds for Material Discrimination in Photon Counting CT

Thomas O’Donnell1, Venkatesh Mani2, Rabee Cheheltani3, David Cormode3, Zahi Fayad2 1Siemens Healthcare, 2TMII Ichan School of Medicine at Mount Sinai, 3Dept. of Radiology University of Pennsylvania

Introduction Photon counting CT (PCCT) is an emerging imaging technology in which individual photons in an x-ray flux are measured using direct converter detectors and categorized via energy thresholds. This facilitates the simultaneous acquisition of multiple image volumes with differing energy characteristics. However, this potentially presents the user with a very large number of choices for energy threshold combinations. For example, a “quad” PCCT system with 100 possible threshold settings has Choose(100,4)= 950,040 options.

Methods & Results In this paper we describe a computationally tractable means to order, from best (most accurate) to worst (least accurate), threshold combinations for the task of discriminating pure materials within a range of assumed concentrations using the Bhattacharyya Coefficient. We use as input, image volumes containing pure samples of each material over the range of possible thresholds and expected concentrations. First, we create a scatter plot of the Hounsfield Units (HUs) of each pure material from the input image volumes. (See Figure 1 for the process of analyzing a dataset of two materials and two energies). Next, we normalize the vectors of each voxel and calculate the angle describing them. Probability distributions of the angles for each material are then computed by fitting a Gaussian model. The overlap of the two probability distributions reflect the seprarability of the materials. Since the true overlap is computationally intractable, we employ the Bhattacharyya Coefficient to estimate it. We scanned a phantom containing 3 materials (calcium, iodine, and gold nanoparticles) of known concentrations at 140kV at both 84 mAs (Clinical dose) and 160 mAs (Maximal dose) on a Siemens Count. Dual threshold mode was employed multiple times to acquire the threshold pairs: 25/50; 30/55; 35/60; 40/65; 45/70; 50/75; 50/80; 50/85; and 50/90keV resulting a total of 14 unique threshold images of 25,30,35,..90keV. These were combined in all possible ways. A sequential protocol was employed with 3 slices acquired, 4mm slice thickness. We found the best threshold combination for separation of the 3 materials to be: 25 and 85keV (two thresholds) and 25, 70, 90keV (three thresholds) at clinical doses.

Conclusion Our approach for separation of materials may be extended to any number of dimensions. Figure 2 shows the case of 3 materials and three threshold images.

Clinical Relevance PCCT holds the promise of simultaneously acquiring images with more than one contrast agent with lower noise and higher resolution. However, in order to employ this new imaging technique, it is necessary to determine the optimal energy threshold combinations. Our approach facilitates this.

87

φ

Angle φ

Figure 1: Two materials and two threshold images. Left: Scatter plot of gold (green) and iodine (red). Middle: Scatter plot normalized to unit polar coordinates. The angle phi for each voxel is of interest. Right: Gaussian probability distributions of phi for each material. The orange signifies overlap.

Φ 80keV 2 keV

90 keV Φ1

25 keV keV

Figure 2: Three materials and three energy theshold images. Left: Cartoon scatter plot of calcium (blue), gold (green), and iodine (red). Middle: Cartoon scatter plot reduced to unit spherical coordinates. Angles phi1 and phi2 are of interest. Right: Gaussian probability distributions of phi1 and phi2 for each material.

88

Dual-tracer 18F-FDG/18F-NaF PET/MR imaging: feasibility, protocol optimization and quantitative benefits

Nicolas A. Karakatsanis1, Ronan Abgral1,2, Gilles Boeykens1,3, Claudia Calcagno1, Marc Dweck1,4, Philip Robson1, Maria G. Trivieri1, Charalampos Tsoumpas1,5, Zahi A. Fayad1

1Icahn School of Medicine at Mt Sinai, Translational and Molecular Imaging Institute, New York, NY, USA 2University Hospital of Brest, Nuclear Medicine, EA3878 GETBO, Brest, FRANCE 3Academisch Medisch Centrum, Amsterdam, NETHERLANDS, 4University of Edinburgh, British Heart Foundation/University Centre for Cardiovascular Science, Edinburgh, UK 5University of Leeds, Division of Biomedical Imaging, Leeds, UK

Introduction PET imaging can simultaneously utilize multiple tracers to track complementary mechanisms of the same disease in a single subject and, thus, enable multi-parametric and potentially more complete diagnostic assessments. In this study, we are proposing an optimized imaging framework of combined 18F- Fluorodeoxyglucose (18F-FDG) and 18F-sodium fluoride (18F-NaF) tracers. The dual-tracer synthesis and scan time window are optimized, based on dynamic analysis of the measured cocktail pharmaco-kinetic attributes in preclinical and clinical studies. Our aim is to demonstrate the clinical feasibility and potential benefits of FDG:NaF cocktail imaging, particularly for PET/MR lesion detectability and quantification.

Methods & Results Initially, the proposed dual-tracer PET protocol is validated and optimized on two healthy control rabbits (3-5kg). A total dosage of 1mCi/kg was administered for all single- and dual-tracer preclinical studies. First, two individual 18F-FDG PET/CT and PET/MR dynamic scans, both at 0-90min post-injection (p.i.), were conducted on each rabbit, employing Siemens Biograph © mCT and mMR scanners, respectively. The same series was then repeated with a 18F-NaF tracer. Subsequently, each tracer spatiotemporal distribution was evaluated in selected regions of interest (ROIs) in the liver, myocardium, heart left-ventricle and bone tissues. Later, a series of cocktail dynamic 0-90min PET/CT and PET/MR scans were performed on the same cohort of rabbits to enable comparisons of the individual as well as the mixed tracer components time activity curves (TACs) at 9:1, 4:1 and 1:1 FDG:NaF ratios. In addition, a series of three 0-90min head and neck PET/MR dynamic scans were conducted on each member of a cohort of human subjects, involving a) a 10mCi dose of the cocktail FDG:NaF tracer, at the previously optimized ratio, as well as individual b) 5mCi FDG and c) 5mCi NaF dosages. Our quantitative analysis on dynamic rabbit PET/CT and PET/MR data demonstrated for bone tissues a higher PET uptake rate for NaF-only relative to FDG-only acquisitions. On the other hand, the time course of the PET signal in the blood plasma (input function) is not affected by the tracer composition itself, while the uptake in the liver is lower for NaF-only scans. Both individual tracer TACs appear to stabilize after 60min p.i., except from NaF uptake in bone, which continues to rise, thus suggesting an optimal scan window of 60-90min p.i. In addition, the maximum FDG:NaF mixture ratio, for which bone tissues can be segmented for more accurate MR- guided attenuation correction was found to be 4:1. Preliminary evaluation on human cocktail PET/MR data of the dynamic PET uptake on the carotid lumen and vessel walls confirmed the previously optimized scan time window and cocktail components concentration ratio.

89

Conclusion Dual-tracer 18F-FDG/18F-NaF PET/CT and PET/MR imaging can be translated to clinical routine by utilizing a 60- 90min p.i. acquisition time window. Moreover, PET/MR lesion quantification may be enhanced with a cocktail tracer synthesis of 4:1 or less ratio, such that the majority of bone regions are segmented in MR-based attenuation maps from the NaF component of the cocktail PET data.

Clinical Relevance Multi-parametric 18F-FDG/18F-NaF PET/MR imaging is clinically feasible and may potentially enhance molecular diagnostic assessments by exploiting the two provided channels of PET tracer signal for the simultaneous evaluation of inflammation and micro-calcification progress in cardiovascular diseases. Moreover, the NaF component could be further utilized to improve PET quantification in the absence of bone attenuation maps.

Figures

Figure: (a) 60-90min PET rabbit images of different cocktail FDG:NaF synthesis and respective bone TACs. (b) 4-tissue class segmentation for MR-based attenuation correction of PET human data using golden angle radial vibe (GAR vibe) and conventional Dixon MR sequences. (c) Carotid images acquired with various MR sequences and fusion with 60-90min PET data for subject 1, and(d) same for subject 2. (e) FDG/NaF and FDG PET images of subject 1, reconstructed from data at early and late post-injection scan time windows.

90

Assessment and comparison of attenuation characteristics of two flexible MRI coils in order to achieve optimal quantification of PET/MR data.

Authors & Affiliations: Alison N. Pruzan, Audrey Kaufman, Zahi Fayad, Venkatesh Mani Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, USA Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY USA Introduction: In order to accurately quantify PET/MR results, one must take into account the attenuation properties of the coils used. Typically, attenuation maps for fixed coils are built into the system and automatically accounted for during PET/MR imaging. However, flexible MRI coils such as those used for chest or carotid imaging do not have prebuilt attenuation maps incorporated into the system. Some such coils are created in a PET/MR friendly design and do not attenuate significantly and therefore may be used in PET/MR without the need for attenuation correction. During this experiment we sought to check attenuation properties of two such coils by acquiring a CT scan of the coils. Towards this end we evaluated the CT attenuation of two different coils, MR Instrument Coil (MRIC) (Figure 2) and Siemens Body Array Coil (SBAC) (Figure 3) in order to check their attenuation properties. If the coil attenuation is low enough, it is possible that PET/MR images need not be corrected for the effects of these coils. Methods & Results: Two coils (MR Instrument Coil and Siemens Body Array Coil) were scanned on a Siemens Somatom Force CT Scanner (Erlangen, Germany). Coils were positioned on the table, curved around pillows to provide the shape of a torso and taped for stability. Coils were placed one below the other on the table, with the MRIC superior (head direction), and the SBAC inferior (foot direction) of the scanner. Two dual energy acquisition scans were administered The first consisted of an 80kV and 150kV with a tin filter. The second scan was a 100kV and 150kV with a tin filter. All images were then reconstructed. Separate 80KV, 100KV, and 150 KV with tin filter images were first reconstructed. Then a blend of the 80 kV and 150 kV with Sn filter and a blend of the 100 kV and 150 kV with Sn filter images were also reconstructed. Following image acquisition and reconstruction we looked at the attenuation properties of all the images (Figure 1). Reconstructed axial images were evaluated on OsiriX by drawing uniform square ROIs over the field of view of each image. ROIs were kept constant across all images to ensure uniformity. Once ROIs were propagated and saved across all images, ROIs were exported to Excel. Mean, min, and max values were extracted and analyzed. The mean of the means, and the standard deviation of the means were also calculated. These values were plotted in Prism (Figure 1). MRIC had a wider range of attenuation values, with higher minima and maxima, whereas the SBAC had a smaller range of attenuation values with a lower standard deviation across all scans. Conclusion: This data indicates that although the SBAC Coil had a lower attenuation than the MRIC coil, there is no statistical difference between the two coils. Both coils however did have significant attenuation and cannot be used without atteniuation correction for PET/MR imaging. Clinical Relevance This method is of clinical relevance because by determining which coil causes less error in PET quantification in PET/MR scans, we will be able to produce more accurate PET/MR results in cardiovascular imaging.

91

Figures and tables: Figure 1: Coil Comparison Graphs

Figure 2: MRIC Coil Figure 3: SBAC Coil

92

Feasibility of Vessel Wall Imaging of the Superficial Palmar Arch using 7T and 3T MRI in comparison with Micro Ultrasound

Authors & Affiliations: Alison N. Pruzan, Audrey Kaufman, Claudia Calcagno, Yu Zhou, Zahi Fayad, Venkatesh Mani Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, USA Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY USA Introduction: The evaluation of atherosclerosis in smaller arteries in the hand may be clinically useful and provide valuable insight into the progression of cardiovascular disease, especially in individuals with diabetes and Raynaud’s syndrome. Imaging the vessel wall of such small caliber arteries (2.5mm to 3.1mm) requires very high spatial resolution and the 7 Tesla would be the ideal tool to acquire these images. We sought to demonstrate feasibility of vessel wall imaging of the superficial palmar arch using 7T and 3T MRI in comparison with very high frequency micro ultrasound. Methods & Results: Four subjects (ages 22-50) were scanned on a micro-ultrasound system with a 45 MHz transducer (Vevo2100, VisualSonics). The Vevo Imaging station was used for mounting the transducer and for stabilizing the position of the hand. The subject was seated during the scan with the hand in supine position with slight rotation toward neutral. Padding was also used under the hand and arm for stabilization and comfort purposes (Figure 1). With this positioning, images in B mode, Power mode, Doppler mode and M mode of the superficial palmar arch were obtained. We obtained measurements of average peak Doppler velocity, intima media thickness, wall thickness, lumen diameter, and total vessel diameter. Subjects’ hands were then imaged on a 3T clinical MR scanner (Siemens Biograph MMR) using an 8 channel special purpose phased array carotid coil (Figure 1). Lastly, subjects’ hands were imaged on a 7T clinical MR scanner (Siemens Magnetom 7T Whole Body Scanner) using a custom built 8 channel transmit receive carotid coil. (Figure 1). Subjects were imaged in a head-first prone position with hand extended above the head. The imaging protocols between 3T and 7T were matched as closely as possible. Total scan time was approximately 20 minutes each. We acquired a localizer, a 3D time-of- flight (TOF) MR angiography sequence followed by a 3D T2 weighted SPACE sequence with 0.6 mm isotropic resolution in all dimensions. (Table 1) Furthermore, we also measured wall thickness, lumen and outer diameters from the MR images. Both Micro-Ultrasound and MRI images were then subjectively analyzed for image quality. Three readers rated the images on a score of 1-5 with 1 being poorest and 5 being the best. Results from the 3 readers were averaged. Criteria used for subjective evaluation were the overall image quality, visualization of the vessel wall, adequate flow suppression and absence of artifacts. 7 Tesla trended to have better visibility of the vessel and its wall, but there was no significant difference between the three methodologies (Figure 4). Conclusion: Results of this preliminary study indicated that vessel wall imaging of the superficial palmar arch was feasible with a Whole Body 7T MRI with subjective evaluations indicating that the image quality obtained at 7T was superior to both 3T MRI and micro-ultrasound. This method may be of clinical relevance in specific disease conditions such as diabetes. The 3D SPACE sequence with isotropic voxels allowed multi-planar reformatting of images and allows for less operator dependent results as compared to ultrasound imaging. Both a larger sample size and recruitment of individuals at risk for atherosclerosis are needed for follow up studies. Clinical Relevance: With this study we aim to portray the feasibility of non-invasive techniques in the screening of peripheral vascular disease such as diabetes and Raynaud’s syndrome. Furthermore, acquiring data using MRI has the benefit of being a less operator dependent modality, and therefore potentially more accurate.

93

Figures and tables Figure 1: Imaging Setups Figure 4: Subjective Data Analysis

Figure 2: MRI Images (3T, 7T)

Figure 3: Ultrasound Images (B Mode, Doppler, M Mode)

94

Motion Averaged MR-Based Attenuation Correction for Coronary 18F-Fluoride Hybrid PET/MR

Authors & Affiliations Philip M. Robson1, Marc R. Dweck1, Nicolas A. Karakatsanis1, Maria Giovanna Trivieri2, Ronan Abgral1, Johanna Contreras2, Umesh Gidwani2, Jagat P. Narula2, Valentin Fuster2, Jason C. Kovacic2, Zahi A. Fayad1 1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY 2Department of Cardiology, Icahn School of Medicine at Mount Sinai, New York, NY

Introduction Recently, 18F-fluoride PET/CT has been shown to identify micro-calcification in atherosclerotic plaques associated with a recent myocardial infarction [1]. Hybrid PET/MR scanners offer the ability to investigate active coronary atherosclerosis while reducing the radiation dose compared to PET/CT allowing repeated and longitudinal studies. PET image reconstruction requires knowledge of the PET-photon attenuation of the object in order to produce accurate images of PET tracer activity. The current standard approach for MR-based attenuation correction (MRAC) is breath-hold volumetric imaging to freeze motion of the chest and abdomen. For imaging the heart, mismatch of the anatomy due to respiration between PET and MRAC data may cause artifacts. In this work, we propose mapping attenuation using a free-breathing golden-angle radial gradient echo sequence and compare the PET images produced with the novel and standard breath-held approach.

Methods & Results Six patients with diagnosed cardiovascular disease or risk factors were imaged using a Siemens Biograph mMR. PET and MR data were acquired simultaneously between 40 and 90 minutes after injection of 10 mCi 18F-sodium fluoride [1]. The standard approach reconstructs PET data using an end-expiration, breath-hold, 3D-VIBE, MRAC map (resolution 4.1 x 2.6 x 3.1 mm3, scan time 19 s). Our motion-averaged approach used an MRAC map derived from a free-breathing, golden-angle radial (GAR) stack-of-stars sequence (3D-GAR-VIBE) [2] (resolution 3 x 3 x 3 mm3, scan time ~7 min). In addition, GAR-VIBE- MRAC data were self-gated according to center-of-k-space intensity [3] (as in Siemens WIP 793) to reconstruct an end- expiration GAR-VIBE-MRAC volume. Finally, a breath-held-VIBE-MRAC was acquired at end-inspiration. GAR-VIBE-MRAC maps were segmented into single soft tissue and background air. GAR-VIBE pixel values were plotted as a histogram. A clear peak at zero was discernible from image values in all images (Fig. 1). The first trough was used to automatically segment soft-tissue from background, to be independent of user interaction to find the noise level. For comparison, in the breath-held-VIBE-MRAC maps, both soft tissue and fat were reassigned as soft tissue and lung was assigned as background. PET images were reconstructed offline with the same emission data using each of the four different MRAC maps using e7_tools (Siemens) . Images were analyzed for image quality by an expert panel (blinded at time of image scoring) to assess the presence of attenuation correction image artifacts and scored on a 0-4 scale. Average scores for each MRAC method, and paired t-tests were used to compare methods.

Attenuation-corrected PET images for each MRAC method are shown in Fig. 3. Mismatch between PET signal and attenuation maps was typical with both end-expiration and end-inspiration breath-hold -VIBE-MRAC producing artifacts in the heart/lung and lung/liver boundaries. Using the proposed motion-averaged GAR-VIBE-MRAC these were significantly reduced (p = 0.002) compared to end-expiration DIXON-VIBE-MRAC (Fig. 4). In one patient who had experienced a myocardial infarction, increased 18F-fluoride uptake is evident and co-localized to the left anterior descending artery (gadolinium contrast- enhanced MRA) corresponding to the infarct territory on delayed contrast enhanced MRI (Fig. 5).

Conclusion Motion-averaged MRAC is necessary for artifact-free PET images in the heart. Using the novel golden-angle radial MRAC approach is superior to the standard breath-hold DIXON-VIBE-MRAC approach. Finally, we were able to identify increased 18F-fluoride uptake in the coronary artery plaque of a patient who had recently had a myocardial infarction.

Clinical Relevance 18F-NaF PET/MR is a promising new technology for assessing active micro-calcification in atherosclerotic disease in the coronary arteries. Importantly, the reduced radiation exposure per scan (~5 mSv) compared to PET/CT permits repeated scans for clinical follow up and disease progression/regression studies.

95

Figures are eliminated using the motion-averaged free- breathing GAR-VIBE-MRAC approach.

Fig. 1. Histogram of GAR-VIBE image pixel intensity. The first trough (arrow) is used as a threshold to automatically segment background at low values Fig. 4. Average scores assigned to PET images by from soft-tissue at higher values. expert panel for each MRAC approach (high scores

represent increased artifact). Poor scores are found for both end-expiration and end-inspiration breath- hold approaches. Motion-averaged free-breathing GAR-VIBE-MRAC shows significantly better scores than breath-hold DIXON-VIBE. No significant difference is seen between GAR-VIBE and end- expiration gated GAR-VIBE.

Fig. 2. MRAC attenuation correction maps for A) standard end-expiration breath-hold DIXON-VIBE, B) end-inspiration DIXON-VIBE, C) motion-averaged golden-angle radial VIBE, D) end-expiration GAR- VIBE. Artifacts in the MRAC of the DIXON method are seen on the heart/lung boundary and in the bronchi (arrows).

Fig. 5. PET image fused with gadolinium contrast- enhanced MRA (left) shows increased 18F-fluoride signal co-localized with a left anterior descending coronary artery plaque (arrow), corresponding to the

infarct territory seen on delayed contrast enhanced Fig. 3. Typical attenuation corrected PET images MRI (right) in a patient with recent myocardial with MRAC maps: A) standard end-expiration infarction. breath-hold DIXON-VIBE, B) end-inspiration DIXON- VIBE, C) motion-averaged golden-angle radial VIBE, D) end-expiration GAR-VIBE. Artifacts in the PET image at the heart/lung boundary and in the bronchi Acknowledgments This work was supported by NIH grant R01 HL071021 References: [1] Joshi NV et al. Lancet 2014;383:705-713. [2] Chandarana H et al. Invest Radiol. 2013;48(1):10-6. [3] Grimm R et al. Med Image Comput Comput Assist Interv 2013;16:17-24. 96

High Frequency Ultrasound Imaging Methods to Evaluate Carotid Arteries

Yu Zhou, Lei Feng, Johnny Ng, Cheuk Ying Tang Translational and Molecular Imaging Institute at Mount Sinai Medical Center

Introduction Carotid artery disease may not show signs or symptoms until it severely narrows or is completely occulded. (https://www.nhlbi.nih.gov/health/health-topics/topics/catd/signs). Atherosclerotic plaques is a significant cause of ischemic stroke, a leading cause of death and disability worldwide. We used high frequency ultrasound imaging to explore some fast and low cost screening carotid methods.

Methods & Results We used Vevo2100 micro-ultrasound system ,40MHz transducer ( MS-550D ). Ten wild type mice (BL57/6J) underwent carotid micro-ultrasound with B mode, M mode, color Doppler , Pulse Doppler and three-dimensional ( 3D )system. These imaging methods will be helpful for pre-clinical research, clinical diagnosis and treatment evaluation. 1. Two Dimensional Imaging We used 40 MHz transducer to scan the mouse carotid. One can see the carotid vessel and its walls. The resolution that can be obtained is around 40µm Figure 1. 2. M mode Carotid Imaging We can use high frequency 40MHz transducer to scan the carotid and measure the carotid systolic and diastolic diameter Figure 2. 3. Blood Dynamic Information with Pulse Doppler (PW) System We scanned the Common Carotid and Internal Carotid with Pulse Doppler System, measured speed and collected blood dynamic information Figure 3 . 4. Two Dimensional (2D) Color Carotid Imaging We scanned the Carotid with Color Doppler System. The carotid blood color imaging will give us more detail blood dynamic information Figure 4. 5. Three Dimensional (3D) Carotid Imaging We also scanned the carotid three dimensional structure with Vevo2100 system. These will be used for different angle to diagnose the carotids. Figure 5 & 6.

Conclusion High frequency ultrasound can be used to get high resolution two-dimensional (2D) and three-dimensional(3D) anatomical and functional images of the carotids of rodents.

Clinical Relevance These high frequency ultrasound imaging methods will eventually be helpful for more accurate clinical diagnosis. and fast screening of patients. The goal is to explore some fast and cheap carotid screening methods to provide doctors with multiple information to save life.

97

Figures and tables

Figure 1 Carotid B mode image Figure 2 Carotid M mode image

Figure 3 Carotid PW mode images Figure 4 Carotid 2D blood color image

Figure 5 Carotid 3D image Figure 6 Carotid 3D blood color image

98

Abstracts Selected for Poster Presentation Nanomedicine

99 100

Design and development of nanomedicines to treat atherosclerosis: a cross platform head-to-head theranostic study Amr Alaarg1,2,3, Carlos Pérez-Medina1, Jun Tang4, Francois Fay1, Yiming Zhao1, Brenda L. Sanchez-Gaytan1, Thomas Reiner4, Zahi A. Fayad1, Robbert J. Kok2, Josbert M. Metselaar3, Willem J.M. Mulder1, Gert Storm2,3 1Translational and Molecular Imaging Institute, Mount Sinai School of Medicine, New York, NY 10029, USA 2Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, Utrecht, 3584CG, Netherlands 3Targeted Therapeutics section, Department of Biomaterials Science and Technology-MIRA Institute, University of Twente, Enschede, 7522NB, Netherlands 4Department of Radiology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10065, USA Introduction: Atherosclerosis is a chronic inflammatory disease of the large arteries and a leading cause of death worldwide. Macrophages are key players in the progression of atherosclerosis and are a compelling target for disease management [1]. Statins, HMG-CoA reductase inhibitors, exhibit anti-inflammatory and anti-proliferative pleiotropic effects [2]. Using a nanomedicine approach these effects can be amplified [3]. Here, we systematically study three different statin-loaded nanoparticles in atherosclerotic apoE -/- mice. Since the nanomedicines also contain diagnostic probes in addition to the encapsulated drug, we can use powerful imaging modalities like positron emission tomography with computed tomography (PET/CT) and near-infrared fluorescence (NIRF) imaging to follow the fate of the theranostics nanoparticles. Methods: Simvastatin loaded high-density lipoprotein ([S]-rHDL), polymeric nanoparticles ([S]-PN), and liposomes ([S]-lip.) were developed. The three formulations were characterized by transmission electron microscopy (TEM), dynamic light scattering (DLS) for mean size and polydispersity (PDI), as well as for drug entrapment efficiency (EE%) by high performance liquid chromatography (HPLC). The effect of the three formulations on RAW 264.7 macrophage viability was evaluated in vitro. For multimodal imaging, Cy5.5 phospholipid (for NIRF imaging) and desferrioxamine (DFO)-phospholipid (for 89Zr-labeling and PET imaging), were included in the [S]-rHDL and [S]-lip. formulations, while for [S]-PN, Cy5.5 and DFO were covalently conjugated to the polymeric nanoparticles. In apoE -/- mice, the nanoparticles were injected i.v. 24 hours before performing in vivo PET/CT. The radioactivity and dye distribution were assessed by gamma counting, autoradiography and NIRF imaging ex vivo. Results: The [S] formulations were successfully prepared and had average sizes < 100 nm, with a narrow PDI, and EE > 60% (Fig.1B and C). [S]-PN shows a more potent effect on RAW 264.7 cell viability with IC50 of 5 µM, compared to IC50 of 10 µM for [S]-HDL and >25 µM for [S]-L (Fig. 1D). Distinct biodistribution profiles for the three nanoparticles were observed by PET/CT imaging, which was corroborated by ex vivo gamma counting (Fig 1E and F). [S]-PN accumulates to a higher degree in spleen and liver compared to [S]-lip. and [S]-rHDL, while the latter [S]- rHDL accumulates to a higher extent in the kidneys. [S]-lip. shows slightly higher concentration in blood at 24 hours. Radioactivity concentration in the aortas was similar for the three nanoparticles. Ex vivo NIRF imaging and autoradiography demonstrated co-localization of 89Zr with cy5.5 signal in the aortas (Fig. 1G and H). Conclusion We have developed three [S]-loaded nanoparticle platforms and labeled them for imaging with PET and NIRF. This allowed us to non-invasively visualize their distinct biodistribution profiles and to assess their plaque targeting ability. Studies are ongoing to evaluate the impact of platforms on the therapeutic outcomes.

101

Clinical Relevance Designing and developing novel targeted therapeutics can circumvent the side effects of small molecule drugs while maximizing their therapeutic activities. We hereby develop novel simvastatin platforms with paying careful attention to the drug-nanocarrier. Comparing the different platforms will provide insights about which platform can be taken for further pharmaceutical development under Good Manufacturing Practice (GMP) conditions, for validation in atherosclerotic patients with advanced inflammatory plaques.

1.Schulz & Massberg . Sci. Transl. Med. 2014 2. Tang et al. Sci. Adv 2015 2.Mulder et al. Sci. Transl. Med. 2014

Figure

Figure 1. Development and multimodal imaging of three Simvastatin [S]-nanoparticle platforms. A) illustrative representation of nanoparticles, B) TEM images, C) characterization of size, PDI and drug content, D) cell viability of murine RAW 264.7 macrophages incubated for 48 h with the nanoparticles, E) representative images of in vivo PET/CT imaging at 24 h p.i., F) biodistribution of the nanoparticles, assessed by gamma counting ex vivo at 24 h p.i. G) autoradiography and H) NIRF imaging of dual labelled nanoparticles in atherosclerotic aortas ex vivo at 24 h p.i.

102

Multimodal in vivo evaluation of a nanoparticle with a surface-converting coating.

Authors & Affiliations Francois Fay1, Line Hansen1,2, Stefanie J. C. G. Hectors1, Brenda Sanchez-Gaytan1, Yiming Zhao1, Jun Tang1,3, Jazz Munitz1, Amr Alaarg1, Mounia S. Braza1, Claudia Calcagno1, Gustav J. Strijkers4, Zahi A. Fayad1, Carlos Pérez- Medina1 and Willem J.M. Mulder1,5. 1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA; 2Interdisciplinary Nanoscience Center, Aarhus University, Aarhus, Denmark; 3Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA; 4Biomedical Engineering and Physics, Academic Medical Center, Amsterdam, The Netherlands; 5Department of Medical Biochemistry, Academic Medical Center, Amsterdam, The Netherlands.

Introduction Active targeting of nanoparticles (NP) through surface decoration with tumor specific ligands is a common strategy to enhanced selective delivery of nanotherapy to the site of disease. Disadvantageously, those moieties may also cause elevated NP recognition by the mononuclear phagocyte system and off-target binding. To overcome these limitations, we have developed a new matrix metalloproteinase-2 (MMP2) sensitive surface-converting polyethylene glycol shielding strategy to prevent NP/cell interactions in the bloodstream. Once exposed to MMP2, i.e. when the NPs accumulate within the tumor microenvironment, the polyethylene glycol (PEG) coating will be cleaved. The resulting surface exposure of the targeting moieties facilitates NP receptor specific association with tumor cells.

Methods & Results NPs are composed of a poly(lactic-co-glycolic acid) (PLGA) core and surface-switchable PEG-lipid coating functionalized with RGD. The nano-precipitation synthesis method allows the incorporation of several imaging labels, including radioisotopes (zirconium-89), fluorescent dyes (Cy5, Cy3.5) and magnetic resonance imaging (MRI) contrast generating molecules (Gd-DTPA). Radionuclide labeling was chosen to quantitatively assess NP pharmacokinetics and biodistribution at tissue level with high sensitivity and accuracy. NP tumor distribution was assessed in vivo at high spatial resolution by MRI, whilst fluorescent labels were used throughout this work to enable the visualization of multiple dyes by fluorescence microscopy, near infrared fluorescence (NIRF) imaging, and high throughput flow cytometry. In vitro studies revealed that RGD functionalized NP display a specific interaction with αvβ3 integrin expressing MDA-MB-231 breast cancer cells, and also confirmed the converting PEG shielding’s MMP2 sensitivity. Pharmacokinetic measurements revealed that the PEG-shielded NPs exhibited longer blood half-lives than unshielded RGD functionalized NPs. The improved circulation times directly translated into higher tumor accumulation, while MRI indicated that NPs primarily accumulated within the tumor rim. Finally, flow cytometry experiments revealed a higher number of tumor cells positive for the surfaced switching NPs compare to control NPs.

Conclusion We have developed a hybrid NP platform with a MMP-2 cleavable PEG-lipid corona. Multimodal imaging enabled us to demonstrate that this new enzyme specific surface-converting coating strategy ensures a high cell targeting specificity without compromising favorable NP pharmacokinetics.

103

Figures and tables

In blood circulation Within tumor A B mPEG-MMP2p-DSPE !!

Gd-DTPA-DSA !! 89 Zr-C34-DFO

!! !! PLGA-Cy5.5 RGD-mPEG-DSPE C !! !! MMP2

Liv Kid Lun Spl Tum Mus Bra Hea

Figure 1. Schematic depiction of the integration of multiple Figure 2. NP biodistribution analyses by MRI (A), NIRF (B), contrast agents within the NP. and autoradiography (C).

Figure 3. Schematics indicating how nanoparticle coating modulates cellular specificity.

104

Protein-Engineered Ferrofluid for MRI-monitored Chemotherapeutic Delivery

Lindsay K. Hill1,2,3, Jeffrey Liu2, Teeba Jihad2, Sumayya Vawda2, Youssef Zaim Wadghiri3, and Jin Kim Montclare1,2,4 1Biomedical Engineering, SUNY Downstate Medical Center, Brooklyn, NY 11203 2Chemical and Biomolecular Engineering, NYU Tandon School of Engineering, Brooklyn, NY 11201 3Radiology, NYU Langone Medical Center, New York, NY 10016 4Chemistry, New York University, New York, NY 10003 Introduction Theranostic imaging combines diagnostics and monitored drug therapy, making it promising in cancer treatment. Protein engineering allows for the biological synthesis of therapeutic platforms combinable with inorganic materials serving as imaging probes. Here, we develop a protein-engineered ferrofluid as a magnetic therapeutic delivery system, for use in chemotherapeutic delivery and as an MRI agent. Our system is based on a coiled-coiled protein dubbed Q, maintaining a hydrophobic pore for binding small molecules, such as anti-neoplastic curcumin. Azide-functionalized Q is synthesized via non-natural amino acid incorporation, rendering it capable of azide- alkyne cycloaddition to an alkyne-bearing iron oxide (FeOX)-templating peptide, CMms6, derived from the magnetosome-associated protein Mms6. The resulting complex is a magnetic protein-engineered chemotherapeutic delivery system. Methods & Results Azide-functionalized Q is expressed in methionine auxotrophic E. coli in media supplemented with azide-bearing azidohomoalanine in place of methionine. Q protein is dialyzed into acidic pH, resulting in nanofibers visible with transmission electron microscopy (TEM). Nanofibers are bound to curcumin to form micron-scale fibers then stabilized via chemical crosslinking and visualized under fluorescence microscopy due to the fluorescent nature of protein-bound curcumin. Following fiber stabilization, alkyne-CMms6 is conjugated to Q via azide-alkyne cycloaddition. The co-precipitation of iron salts in the presence of Q-CMms6 and NaOH reduction yields templated iron oxide nanoparticles. TEM and fluorescence microscopy of protein revealed fibers from the nanoscale (~600 nm diameter), to mesoscale, (~35 µm diameter) before and after curcumin binding, respectively. Nanoparticle features were assessed with TEM, showing cuboidal FeOx particles 7 nm in diameter. Ferromagnetism was confirmed with magnetic manipulation and relaxometry was performed at 7-Telsa using multi-echo schemes to acquire R2 (1/T2) and R2* (1/T2*) relaxation maps. Relaxometry confirmed a 3-to-3.4-fold increase in relaxation rate R2 and a 10.7- to-21.4-fold R2* increase compared to a buffer control, even with observed ferrofluid settling, showing promise for MRI monitoring.

Conclusion By taking advantage of a biologically synthesized coiled-coil protein and bio-inspired FeOx templation, we have created a magnetically-functionalized drug-carrying ferrofluid. Initial ferromagnetization and preliminary MRI studies suggest potential for magnetically-driven drug delivery and visualization and tracking via T2/T2*-weighted imaging. Clinical Relevance Progress in theranostics development will likely have a tremendous effect on rapid detection and treatment of various cancer types. In translating this work to the clinic, however, is it critical to consider the safety of these

105 agents. The proposed agent combines a protein-based vehicle and FeOx nanoparticles, both well-tolerated in vivo, and should yield a biocompatible theranostic ferrofluid.

Figures and tables

Curcumin-bound Q protein fibers templated to iron oxide are visualized under fluorescence microscopy demonstrating an average fiber diameter of 23.9 µm (14- 33.6 µm, n=11) (a). Brightfield microscopy revealed fibers coated in iron oxide particles due to conjugation of the iron-templating peptide CMms6 (b). This complex is also characterized using TEM to reveal cuboidal nanoparticles with an average diameter of 6.6 nm (4.6 nm-10.2nm, n=20) (c). Energy Dispersive X-ray Analysis (EDAX) suggests the presence of iron oxide particles spatially overlapped with protein fibers based on elemental analysis (d). Relaxometry comparing diluted FeOX- bound fibers to a phosphate buffer control shows a 3-to-3.4-fold increase in R2 (e) and a 10.7-to-21.4-fold increase in R2* (f).

106

Rapid Quantification of Gadolinium in Nanoparticles by Time-Resolved Fluorescence Lindsay K. Hill1,2,3, Stewart Russell4,5, Dung Minh Hoang2, and Youssef Zaim Wadghiri2 1Biomedical Engineering, SUNY Downstate Medical Center, Brooklyn, NY 11203 2Radiology, NYU Langone Medical Center, New York, NY 10016 3Chemical and Biomolecular Engineering, NYU Tandon School of Engineering, Brooklyn, NY 11201 4Thayer School of Engineering, Dartmouth College, Hanover, NH 03755 5Mechanical Engineering, The City College of New York, NY 10031 Introduction The use of Gadolinium (Gd)-based MRI contrast agents (GBCA) in dynamic, such as perfusion MRI (pMRI), scans provides valuable information about tissue microenvironment and function. However, accurate concentration-time curves are required to quantitatively characterize diseased tissue, monitor therapy, and provide drug mechanism insight. We previously developed a technique to quantify a GBCA using carbostyril 124 (cs124)-sensitized DTPA via energy transfer to TbDTPA on a fluorescence plate reader. This work examines whether our method can be extended to multivalent Gd complexes. Methods & Results Amide chemistry was used to conjugate cs124 to DTPA-DPPE and Gd was chelated in water, yielding cs124-GdDTPA-DPPE. The absorption of cs124 at 330 nm was used to quantify Gd. Dialyzed product size distribution was assessed via Dynamic light scattering (DLS). Time-resolved fluorescence of cs124- GdDTPA-DPPE in the presence of TbDTPA, using excitation at 330 nm and emission at 480 nm, was integrated from 600 µs-2000 µs. The collision model of energy transfer was fit to fluorescence data to determine system sensitivity. T1-weighted MRI and relaxometry were acquired on a 7-Telsa Bruker system using phantoms of cs124-GdDTPA-DPPE particles and clinical Gd-DTPA (Magnevist). DLS showed a monodisperse distribution with an average 575 nm radius, consistent with liposomal structure expected from this lipid. The starting Gd concentration used for fluorescence studies was 35 µM. The limit of detection was 67 nM and comparable to ICP-MS. T1-weighted MRI showed that our multivalent cs124-based T1-agent achieved a signal comparable to that of 10x higher concentration Magnevist. Relaxometry confirmed a 26.5x amplification in r1 and a 1.04 r2/r1 ratio. Conclusion Incorporating cs124 into multivalent GBCAs proved effective in maintaining nanomolar quantification of Gd using a readily available spectrophotometer. The cs124 conjugation did not hinder the agent’s r1 relaxivity, suggesting its potential use in other novel GBCA of various sizes. This system may significantly ease the characterization and optimization of novel in-house MR agents. Clinical Relevance This sensitive and accessible technique allows for optimization of novel GBCAs and, we hypothesize, the quantification of these agents upon vascular injection to calibrate the arterial input function in preclinical pMRI studies of tumors, strokes, and other vascular diseases. Improved ease and accuracy of quantification is important for agent development, but its quantification in the blood is critical for such dynamic studies, aimed at understanding the changes in blood flow in diseased states and responsiveness to therapy.

107

Figures and tables

(a) Chemical structure of Gd-chelated cs124-DTPA-DPPE, where fluorescent antenna carbostyril 124 (cs124) is in orange. (b) Schematic of the fluorescence resonance energy transfer between cs124-GdDTPA-DPPE, the energy donor, and Terbium-DTPA (TbDTPA), the acceptor, to quantify the Gd concentration of our agent. Dilutions of cs124-GdDTPA-DPPE particles were added to a 96-well plate with constant TbDTPA. Briefly, cs124 absorbs light (green arrow) and bound gadolinium’s heavy atom effect results in intersystem crossing 5 where energy is lowered to the T1 state and transferred by collision to the excited state D4 of Tb. Time- resolved detection of Tb fluorescence (red arrows) is proportional to the concentration of sensitized Gd. (c) Calibration curve of the cs124-GdDTPA-DPPE construct determined by time-resolved fluorescence emission of TbDTPA. Points above 3x standard deviation of the noise were fit to a collision model (bold dashed line, p < 5x10-5). The resulting sensitivity spanned from the millimolar to nanomolar concentrations with a 67nM limit of detection. (d) Phantom layout and T1-weighted image of either HEPES, HEPES-diluted Magnevist (black text), or HEPES-diluted cs124-GdDTPA-DPPE particles (red text and circle). The signal from 0.05 mM cs124-GdDTPA-DPPE is qualitatively comparable to that of 0.5mM Magnevist. Table 1 Relaxivity values, r1 and r2, for Magnevist and cs124-GdDTPA-DPPE as determined using R1 (1/T1) and R2 (1/T2) phantom maps at 7-Tesla. Each relaxation curve contained at least four concentrations of each agent. The ratio r2/r1 corresponds to the effectiveness of the construct as a T1-agent, with the shortest value being optimal. 108

Multimodal PET imaging of high-density lipoprotein's trafficking in multiple atherosclerosis models

Carlos Pérez-Medina1, Tina Binderup2, Mark Lobatto3, Samantha Baxter1, Claudia Calcagno1, Seigo Ishino1, Thomas Reiner4, Jason S. Lewis4, Zahi A. Fayad1 and Willem J.M. Mulder1 1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2 Clinical Physiology, Nuclear Medicine, PET and Cluster for Molecular Imaging, University of Copenhagen, Denmark 3 Department of Vascular Medicine, Academic Medical Center, Amsterdam, The Netherlands 4 Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, NY, USA ‖ Introduction High-density lipoprotein (HDL) is a natural nanoparticle that interacts with atherosclerotic plaque macrophages to facilitate reverse cholesterol transport. HDL-cholesterol concentration in blood is inversely associated with risk of coronary heart disease and remains one of the strongest independent predictors of incident cardiovascular events. Our aim was to develop and validate a non-invasive imaging tool to visualize HDL’s in vivo behavior by positron emission tomography (PET), with an emphasis on its plaque targeting abilities, in three preclinical atherosclerosis models.

Methods & Results Discoidal HDL nanoparticles were prepared by reconstitution of its components apolipoprotein A-I (APOA1) and the phospholipid DMPC. For radiolabeling with Zirconium-89 (89Zr), the chelator DFO was introduced by conjugation to APOA1 –for 89Zr-APOA1-labeled HDL, 89Zr-AI-HDL– or as a phospholipid-chelator (DSPE-DFO) –for 89Zr-phospholipid-labeled HDL, 89Zr-PL-HDL–. Radiolabeled HDLs’ biodistribution and plaque targeting was studied in established murine, rabbit and porcine atherosclerosis models by PET/CT and, for rabbits, also by PET/MRI. Ex vivo validation was conducted by radioactivity counting and autoradiography. In mice and rabbits, fluorescent analogs were coinjected to allow near infrared fluorescence imaging analysis. We observed distinct pharmacokinetic profiles for the two 89Zr-HDL nanoparticles, 89Zr-AI-HDL showing a significantly longer blood circulation time. Similar radioactivity distribution patterns were found in all three animals for each nanotracer. Both 89Zr-AI-HDL and 89Zr-PL-HDL accumulated in the atherosclerotic vessel wall as well as in kidneys, liver, spleen and bone marrow with some marked quantitative differences in radioactivity uptake values. Significantly increased radioactivity concentrations were measured in rabbit atherosclerotic aortas compared to controls by PET/CT for 89Zr-PL-HDL (0.31 ± 0.10 vs 0.16 ± 0.03 g/mL, P < 0.05), and by gamma counting for both 89Zr-AI-HDL (0.025 ± 0.009 vs 0.007 ± 0.001 %ID/g, P < 0.05) and 89Zr-PL-HDL (0.011 ± 0.002 vs 0.003 ± 0.001 %ID/g, P < 0.05) at 5 days p.i. PET/CT and PET/MR images were in excellent agreement. In pigs, atherosclerotic lesions in the femoral arteries showed high radioactivity accumulation at 48 h p.i. of 89Zr-PL-HDL

Conclusion & Clinical Relevance 89Zr labeling of HDL allows studying its in vivo behavior by non-invasive PET imaging, including visualization of its specific accumulation in the atherosclerotic vessel wall. The different labeling strategies provide insight on the pharmacokinetics and biodistribution of HDL’s main components, i.e. phospholipids and APOA1, and could be of great value to assess HDL-cholesterol trafficking in the context of cardiovascular disease.

109

Figures and tables

A B 60 Control 1.11.1 Control Athero 0.8 Athero 40 0.5 20 0.5 20 0.20.2 %ID/g 10 %ID/g 0.10.1 * Zr-AI-HDL Zr-AI-HDL 89 89 00 0.00 Ki Li Sp Lu Mu Ki Li Sp Lu Mu 120 h 120 h liver 24 h Liver lung Spleen Lungs Muslce kidney spleen muscle Kidneys Blood

2020 0.40.4 Control * Control Athero Athero 15 * 0.30.3

1010 0.20.2 Zr-PL-HDL %ID/g

Zr-PL-HDL * %ID/g 5 89 0.1

89 0.1 * 0 0 0.00 Ki Li Sp Lu Mu 120 h 120 h Ki Li Sp Lu Mu 24 h Liver Spleen Lungs Muscle Kidneys liver lung C D kidney spleen muscle Control Copy of PL_HDL_pig_SUV 0.6 Athero 6 6 1 1.00 -2 * * 20.02-2 0.75 0.4 4 4 0.5 0.50

0.01 SUV 0.2 1 2 2 0.2 0.25 %ID/g x 10 SUV (g/mL) %ID/g x 10 0.0 0 0 0.00 0 0 0 0.00 PET/CT γ-count. Ki Li Sp Lu Mu 1 2

liver lung 48 h kidney spleen muscle

Figure 1. A) Apoe-/- mice 3D-rendering PET/CT fusion images and radioactivity distribution in selected tissues for 89Zr-AI-HDL (top) and 89Zr-PL-HDL (bottom) at 24 h p.i. [Ki: kidney; Li: liver; Sp: spleen; Lu: lungs; Mu: muscle] (n ≥ 4 per group). B) PET/CT and PET/MR fusion images of rabbits with atherosclerosis at 5 days p.i., and radioactivity distribution in selected tissues for 89Zr-AI-HDL (top) and 89Zr-PL-HDL (bottom) at 5 days p.i. [Ki: kidney; Li: liver; Sp: spleen; Lu: lungs; Mu: muscle] (n = 4 per group). D) Comparison between SUVs measured from PET/CT images and radioactivity concentration in aortas of rabbits with atherosclerosis and wild type controls at 5 days p.i. of 89Zr-PL- HDL (bottom) (n = 4 per group). D) Whole-body maximum intensity projection PET image of a pig injected with 89Zr-PL-HDL at 48 h p.i. (arrows indicating lesions), and radioactivity distribution in selected tissues for 89Zr-PL-HDL at 48 h p.i. determined by gamma counting (n = 3). * P < 0.05.

110

A microfluidics-based bottom-up approach for the synthesis of nanostructures for molecular imaging

Authors & Affiliations Brenda L. Sanchez-Gaytan1, Yiming Zhao1, Francois Fay1, Zahi A. Fayad1, and Willem J. M. Mulder1,2 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, 10029 USA. 2 Department of Vascular Medicine, Academic Medical Center, Amsterdam, 1105 AZ The Netherlands.

Introduction The number of available nanomedicine-based imaging alternatives have had a large increase due to improved synthesis approaches capable of producing highly-reproducible homogenous nanoparticles at large scale. Simple and cost-effective methodologies are indispensable for large-scale in vivo trials, however, typical methods to synthesize self-assembled nanoparticles frequently involve ineffective and complicated procedures. Among the plethora of platforms, lipid-based nanoparticles, such as nanoemulsions and the natural-occurring high-density lipoprotein (HDL), are of special interest. Nanoemulsions (oil nanodroplet cores surrounded by a lipid corona) are usually synthesized through top-down approaches,1 which restricts size and modularity. HDL on the other hand, structurally similar to an oil in water emulsion, has inherent target-specific characteristics, and can also be modified with diagnostic moieties (e.g. paramagnetic, fluorescent) or hydrophobic payloads (e.g. nanocrystals).2

Methods & Results A bottom-up microfluidics-based approach was used to synthesize the nanoparticles. Nanoemulsions were generated by injecting ethanolic solutions containing oil, lipids and imaging agents (i.e., nanocrystals, dyes and labeled lipids) into the microfluidics module where they were swiftly mixed with PBS. The procedure yielded nanoparticles with imaging capabilities in the core and/or the corona. The ratios of oil/lipids as well as the flow rates and the viscosity of the lipids were found to have a profound impact on the size and overall quality of the nanoparticle. Sizes varied from 30 to 150 nm, with high flow rates generating the most monodispere batches. Loading of other components such as nanocrystals and other imaging agents in both the core and the corona was very efficient and highly improved compared to top-down synthesis methods. μHDL with different phospholipid compositions and loadings was prepared as well. For this, ethanolic solutions containing phospholipids and imaging agents were mixed with apoliporotein AI in PBS. Upon injection of the components, μHDL forms instantaneously and the resulting nanoparticle has the same structural and biological properties (e.g., size, morphology, bioactivity) as native HDL. In addition, imaging agents could also be easily incorporated.

Conclusion Microfluidic-based technology was employed to self-assemble synthetic lipid-based nanoparticles with molecular imaging capacities. The ease of preparation allows the production of highly reproducible nanoparticle batches with modulation of structure and composition not possible through typical top-down synthesis approaches. 1. Delmas et al. Langmuir 2011. 27. 5. 1683 2. Duivenvoorden et al. Nat Commun. 2014. 5. 3065

Clinical Relevance The microfluidics-based method presented offers a reproducible and modular approach for the large-scale synthesis of nanostructures containing contrast agents that can be used for molecular imaging.

111

Figures and tables

Figure 1. (a) Transmission electron micrograph (TEM) of nanoemulsions and (b) TEM of nanoemulsions incorporating gold nanocrystals, scale bars 250 nm. (c) Size dependency of nanoemulsion on oil to lipids ratio and flow rate. Uptake of μHDL by macrophage cells in vitro (d) Fluorescent micrographs of cells incubated for 2 h with rhodamine-labeled μHDL and rhodamine-labeled control DSPC/DSPE-PEG NP (e) scale bar 50 um. (f) TEM of HDL nanoparticles, scale bar 100 nm.

112

A novel tracer for non-invasive atherosclerotic plaque phenotyping by PET/MR imaging in experimental models of atherosclerosis

Max L. Senders1, Calvin Yeang2, Hannah Groenen1, Francois Fay1, Claudia Calcagno1, Simone Green2, Phuong Miu2, Thomas Reiner3, Joseph L. Witztum2, Zahi A. Fayad1, Willem J.M. Mulder1, Carlos Perez-Medina1 and Sotirios Tsimikas2 1 Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA 2 Sulpizio Cardiovascular Center, Department of Medicine, University of California, La Jolla, San Diego, CA, USA. 3 Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, NY, USA

Introduction Atherosclerosis is the major underlying cause of cardiovascular disease. Oxidation specific epitopes (OSE) are well defined danger-associated molecular patterns that activate inflammatory pathways leading to initiation and progression of atherosclerotic plaques.1 To detect (sub)clinical atherosclerosis, a variety of imaging techniques are available. However, current clinical techniques do not characterize the atherosclerotic plaque well, but merely detect the level of stenosis. In the past decade, molecular imaging has provided new insights in pathophysiology at a cellular and molecular level. It therefore allows phenotyping and identifying features of plaque progression that precede a possible plaque rupture.2 We have developed a new OSE-specific PET imaging probe and evaluated this tracer using a multimodality imaging approach.

Methods & Results Human monoclonal antibody Fab LA25 was generated from a phage display library screened for OSE binding. LA25 was modified with the chelator deferoxamine B and radiolabeled with Zirconium-89 (89Zr-LA25). Pharmacokinetics, biodistribution and plaque specificity were extensively assessed in cholesterol-fed apoE-/-mice and a rabbit model of atherosclerosis. For in vivo imaging, a clinical positron emission tomography with magnetic resonance imaging (PET/MRI) system was used. First, the level of vessel wall inflammation was determined by a 18F-FDG PET/MR scan.3 The next day, after intravenous injection with 89Zr-LA25 or 89Zr-LA24, a non-specific Fab, rabbits were scanned in a dynamic fashion for 1 hour (fig.1A) with simultaneous acquisition of dynamic contrast enhanced MRI (DCE-MRI) to evaluate neovascularization.4 An additional scan was performed 24 hours post injection, after which the animals were sacrificed to assess macrophage content by ex vivo near infrared fluorescence (NIRF). Dynamic PET imaging data revealed an increase in renal signal over time (fig.1B) indicative of renal clearance. The weighted half-life in atherosclerotic rabbits was 2.14h for 89Zr-LA25 and 1.20h for 89Zr-LA24 (fig.1C). After 24 hours, PET/MR imaging (fig.1D) showed a significant higher uptake of 89Zr-LA25 in the abdominal aorta of atherosclerotic rabbits compared with control (P=0.0003, fig.1E), confirmed by gamma counting (P<0.0001, fig.1F) and autoradiography (fig.1G). Uptake of 89Zr-LA24, was significantly lower than that of 89Zr-LA25 in atherosclerotic rabbits (P<0.0001 fig.1F, G). 18F-FDG PET/MR imaging, DCE- MRI and NIRF showed significant higher uptake in atherosclerotic rabbits (P<0.0001, fig.2 A-D) and the weighted half-life for 89Zr-LA25 was 1.31h in healthy control rabbits. Similar results were obtained in mice showing a significant higher uptake of 89Zr-LA25 in aortas using gamma counting (fig. 2.F, P<0.0001), confirmed by autoradiography (fig. 2.G, P<0.0001). Biodistribution in mice was similar to rabbits, indicative of renal clearance with no significant uptake in the brain (fig.2H). The weighted half-life of 89Zr-LA25 and 89Zr-LA24 in apoE-/- mice was 19.09 min and 9.38 min respectively (fig.2I).

Conclusion & Clinical relevance 89Zr-LA25, a novel radiotracer targeting OSE, enables phenotyping of atherosclerosis using PET/MRI. In combination with previously validated techniques we are now able to visualize and quantify inflammation (18F-FDG), neovascularization (DCE- MRI) and OSE, which are key features of a vulnerable plaque. Moreover, this radiotracer could further help characterize the process of atherosclerosis and ultimately serve as a biomarker in a clinical setting to evaluate therapeutic interventions.

References 1. Briley-Saebo, KC, et al. J Am Coll Cardiol. 57, 337–347 (2011). Presentation category, please mark your preference [ ]Cancer 2. Mulder, WJM, et al. Sci. Transl. Med. 6, 239sr1 (2014). 3. Tawakol, A, et al. J Am Coll Cardiol. 62, 909–917 (2013).

113

4. Calcagno, C, et al. NMR Biomed. 28, 1304–1314 (2015). [ x ]Cardiac [ x ]Nano [ ]Neuro

Figure 1. Summation PET/MR fusion images at 20, 40 and 60 min (A), and radioactivity quantification in major organs after injection with 89Zr-LA25 in atherosclerotic rabbits based on PET/MR imaging (B). Pharmacokinetics in atherosclerotic rabbits of 89Zr-LA25 and 89Zr-LA24 (C). PET/MR fusion image 24 hours post-injection (p.i.) (D) and in vivo quantification of aortic uptake in atherosclerotic and control rabbits 24 hours p.i. of 89Zr-LA25 (E). Ex vivo radioactivity quantification concentration by gamma counting (F) autoradiography of aortas from rabbits with atherosclerosis (black), and controls (white) injected with 89Zr-LA25, and from atherosclerotic rabbits injected with 89Zr-LA24 (grey) at 28 h p.i. (G).*** P < 0.0001.

Figure 2. Uptake of 18F-FDG in rabbit abdominal aorta measured by PET/MR imaging (A). DCE- MRI uptake in rabbit abdominal aortas (B). Representation of Cy5 rHDL uptake in abdominal rabbit aortas (C) quantified by NIRF as Total Radiant Efficiency [p/s] / [μW/cm] (D). Pharmacokinetics of 89Zr-LA25 in atherosclerotic and healthy control rabbits, half-life of 2.14 hours and 1.31 hours respectively (E). A-E atherosclerotic rabbits (black) and healthy control 89 89 -/- rabbits (white). Gamma counting and autoradiography quantification of Zr-LA25 (black) and Zr-LA24 (white) in apoE mouse aortas (F, G), and biodistribution of major organs (H). Pharmacokinetics of 89Zr-LA25; radioactivity half-life 19.09 min. and for 89Zr-LA24 9.38 min. in apoE-/- mice (I). *** P < 0.0001. 114

Abstracts Selected for Poster Presentation Neuroimaging

115 116

Frequency Shift Imaging (FSI) for characterization of cells labeled with superparamagnetic iron-oxide nanoparticles

Judy Alper1,2, Francois Fay1, Hadrien Dyvorne1, Priti Balchandani1,3 1. Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai; 2. Department of Biomedical Engineering, City College of New York; 3. Department of Radiology, ISMMS

Introduction: MRI has been a primary tool used for detecting and tracking the location of cells labeled with superparamagnetic iron-oxide nanoparticles (SPIOs). While most SPIO applications have relied on negative contrast sequences (1), positive contrast imaging of the off-resonance SPIO signal provides clear benefits (2, 3), such as reduction of background tissue signal and possible quantification of labeled cells in the targeted organ. Moreover, SPIO imaging at ultrahigh field strengths, such as 7 Tesla (7T), makes it possible to leverage the greater off-resonance sensitivity afforded by higher field strengths to provide quantitative imaging of smaller cell populations. In this work we introduce frequency shift imaging (FSI), a novel acquisition technique that combines efficient interleaved spectrally selective excitations with fast spiral acquisition to perform comprehensive characterization of the magnetic signature of SPIOs in a reasonable scan time. We demonstrate the performance of the FSI sequence for imaging macrophages labeled with SPIOs and compare the novel sequence with standard negative-contrast acquisitions performed at 7T.

Methods & Results: Cell phantom: Mouse tumor macrophages were grown in DMEM media in two T75 flasks. Both flasks were incubated for 4 hours at 37° Celsius, one with 30 mg Fe/mL ferumoxutol (labeled cells), the other without (control cells). Subsequently, the cells were fixed with 4% PFA and four samples (of 1 million cells each) were made by mixing labeled and unlabeled cells at 0%, 25%, 75% and 100% population ratio. The cell samples were embedded in 2% agar gel for MR imaging. Sequence Design: A 15 ms-long, 170 Hz bandwidth, self-refocused RF pulse was designed using the Shinnar-Le Roux (SLR) algorithm (Fig. 1) in order to provide minimum echo time and high spectral selectivity, as described in (2). Center-out 3D stack-of-spiral readouts were implemented to minimize the echo time in order to reduce signal loss due to transverse relaxation. 15 different frequency points (-1400 to 1400 Hz) were sampled at 200 Hz intervals in interleaved fashion. Interleaving frequency sampling in time allows for fast repetition rates (24 ms between consecutive pulses) while providing sufficient signal recovery (TR=360 ms between consecutive excitations of the same frequency band). MR imaging: The stack-of-spiral FSI sequence acquired 30 partitions at isotropic 1.0 mm resolution, using segmented spiral readouts (140 mm field of view, 15 arms per 2D spiral, 5.5 ms per readout). Total acquisition time for interleaved FSI was 10 minutes. In addition, standard negative-contrast sequences were acquired at 1.0 mm isotropic resolution on the same phantom, including a 3D gradient echo (TE/TR 2.4/5.3 ms) and a 2D interleaved spin echo (TE/TR 8.7/1000 ms). The FSI measured multi-frequency spin echo signal displays the dipole pattern of the magnetic field of the SPIOs . The regions along the main magnetic field axis are excited by positive frequencies, while regions perpendicular to the main magnetic field are excited by negative frequencies. This difference in region excitement between positive and negative frequencies is seen in the different orientations of the highlighted dipole patterns. Moreover, lower frequency shifts lead to increased signal and larger excitation patterns, while higher frequency shifts result in the opposite (Fig 2). The residual background signal that appears at 200 Hz is a consequence of the large water linewidth at 7T. Positive contrast is shown to have increased localization and background suppression compared to negative-contrast sequences (Fig 3). The positive contrast technique can provide a framework for quantitative analysis, since it allows for the detection of pattern size, which is visually proportional to the amount of labeled cells.

Conclusion & Clinical Relevance: In this study, we obtained a comprehensive characterization of the SPIO magnetic signature in one acquisition and with fast spiral readout using the novel FSI sequence. We have demonstrated the efficiency of FSI for positive contrast imaging of SPIO labeled macrophages at 7T, as compared to negative contrast methods. Positive contrast imaging suggests a path for quantifying labeled cells in a targeted organ, with significant biomedical applications. For future work, we plan to compare detection limits between 3T and 7T to evaluate the advantages of 7T imaging for

117

SPIOs. We also plan to perform this imaging in ex vivo tissue samples and in vivo animal models. Our final goal is to develop electromagnetic models to quantify labeled cells content in vivo, based on observed multi-frequency patterns.

Figures:

Figure 1: (A-B) waveforms of the self-refocused RF pulses used in this work (continuous line: real part, dashed line: imaginary part). The two pulses are generated to provide opposite echo phase in subsequent acquisition. The spin echo is sampled at the end at the pulse. (C) Bloch simulation showing the frequency dependence of the subtracted spin echo signal.

Figure 2: Multi-frequency spin echo signal measured with FSI on cell phantom imaged at 7T. Positive frequencies excite regions along the main magnetic field axis while negative frequencies excite regions perpendicular to it. Higher frequency shifts lead to reduced signal as well as smaller excitation patterns.

Figure 3: Comparison of gradient echo (left), spin echo (middle) and FSI (right) acquired at 7T on macrophage cultured cell population with 0, 25, 75 and 100% of cells labeled with SPIOs. The FSI image was generated using the sum of signals acquired in all frequency bands. While negative contrast and distortions affect standard gradient and spin echo sequences, FSI leads to highly conspicuous signal. Signal intensity and spatial extent of dipole patterns scale with the ratio of labeled cells.

References: 1. Bulte JW, Kraitchman DL. Iron oxide MR contrast agents for molecular and cellular imaging. NMR Biomed. 2004;17(7):484-99. 2. Balchandani P, Yamada M, Pauly J, Yang P, Spielman D. Self-refocused spatial-spectral pulse for positive contrast imaging of cells labeled with SPIO nanoparticles. Magn Reson Med. 2009;62(1):183-92. 3. Cunningham CH, Arai T, Yang PC, McConnell MV, Pauly JM, Conolly SM. Positive contrast magnetic resonance imaging of cells labeled with magnetic nanoparticles. Magn Reson Med. 2005;53(5):999-1005.

118

Combined MR Perfusion and Diffusion for Differentiation of Post-treatment Changes from Recurrent High Grade Glioma

Authors & Affiliations Belani P, Doshi A, Delman B, Hormigo A, Germano I, Nael K Icahn School of Medicine at Mount Sinai

Introduction: Differentiation of posttreatment changes (PTC) from recurrent tumor (RT) in treated patients with high-grade gliomas (HGG) remains a diagnostic challenge. The purpose of this study was to evaluate diagnostic performance of multiparametric MRI using a combination of MR perfusion and diffusion for distinguishing PTC from RT in patients with HGG.

Methods & Results: From Jan 2013 to Sep 2015, a total of 42 patients with HGG who developed a new enhancing mass after completion of their standard treatment (gross total resection, radiation and temozolomide) were retrospectively evaluated. MRI scans in which enhancing lesions were first identified, were used for image analysis. Volume-of-interest (VOI) of the enhancing lesions were created. Using coregistered images, mean values of the ADC, DCE-derived Ktrans and DSC-derived rCBV were calculated. Statistical analysis was performed by analysis of variance and logistic regression. Receiver operating characteristic (ROC) analysis was performed to determine the optimal parameter(s) and threshold for diagnosis of recurrence vs. PTC. Twenty-nine patients had RT (confirmed by surgical pathology), while 13 patients were identified as having PTC: radiation necrosis, n=6 (confirmed by surgical pathology), pseudoprogression, n=7 (diagnosis made on imaging as the enhancing lesion progressively decreased in size and resolved after initial appearance on multiple sequential MRI exams (mean F/U time 7 months)). Recurrent HGG showed significantly higher rCBV and Ktrans and significantly lower ADC values compared to PTC (Table 1). There was no statistically significant difference in ADC, Ktrans or rCBV mean values between radiation necrosis and pseudoprogression (p > 0.1). Multivariate logistic regression analysis showed significant contribution from rCBV (p=0.01) and Ktrans (p=0.04), but not from ADC (p=0.7) to differentiate PTC from RT. The best discriminative power from an individual classifier was obtained from rCBV at threshold of 2.2 resulting in an AUC of 0.92 with sensitivity/specificity of 90/92% respectively. In a separate model, a combined Ktrans-rCBV classifier resulted in slightly better discriminative power with AUC of 0.98 and odds-ratio of 61 for differentiation of PTC from RT.

Conclusion: Recurrent HGG showed lower ADC, higher Ktrans and higher rCBV in comparison to PTC including both radiation necrosis and pseudoprogression. The combined rCBV-Ktrans had the highest diagnostic performance for differentiation of PTC from RT compared to any individual or combination of other imaging classifiers.

Clinical Relevance: Differentiation of recurrent tumor from posttreatment changes in treated patients with HGG remains challenging on conventional imaging, often posing a diagnostic dilemma leading to uncertainties with associated patient’s anxiety and possibly delays in interventions. Early differentiation between these two is therefore highly desirable for both the patient and the treating oncologist.

119

Figures and tables

Table 1. Mean values, analysis of variance and AUC analysis for differentiation of recurrent tumor (RT) from Posttreatment changes (PTC) using rCBV, ADC and Ktrans

PTC (n=13) RT (n=29) ANOVA AUC/Sensitivity/Specificity ADC (10−6 mm2/s 1360 1150 0.02 0.82/69/84 Ktrans (1/min) 0.06 0.18 0.002 0.88/93/77 rCBV 1.8 3.9 < 0.001 0.92/90/92

120

Towards accurate spinal cord morphometry with in situ phantom-calibrated gradient non-linearity correction

Authors & Affiliations Joseph Allan Borrello1,2,3 , Joo-won Kim2,4 , Mootaz Eldib2,4 , and Junqian Xu2,4,5

1Graduate School of Biomedical Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 2Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 3Mount Sinai Institute of Technology, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 4Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 5Department of Neuroscience, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction Spinal cord cross-sectional area (SCCSA) has recently been proposed as a promising biomarker for spinal cord atrophy in neuroinflamatory or neurodegenerative [1] diseases, with particular value as a clinical endpoint in therapeutic trials [2]. In order to reproducibly and accurately measure SCCSA, however, spatial distortions due to gradient nonlinearity in large FOV spinal cord images must be corrected in a consistent and platform- independent manner. Although gradient non-linearity correction algorithm and implementations exist from scanner vendors, they usually do not take into account site specific, residual gradient nonlinearity error. To address this issue, we fabricated a low profile in situ spatial calibration phantom to optimize Z-gradient non- linearity unwarping for SCCSA quantification

Methods & Results The phantom was designed in SolidWorks (Dassault Systems) consisting of repeating chambers (Fig. 1A), 3D printed out of acrylate polymer and filled with de-ionized water. CT imaging at 0.9mm isotropic resolution (Fig. 1B,C,D) was used to assess structural fidelity. The phantom was placed on top of the spine array close to the middle of the subjects' spine when imaging four subjects on a 3T system (Skyra, Siemens) using neck and spine array coils. A T2-weighted slab-selective fast spin echo sequence (SPACE) with magnetization restoration was used to collect two images, one including the cervical spinal cord and upper thoracic spinal cord (upper scan) and a second including the lower cervical and thoracic spinal cord (lower scan) after a 126 mm table advancement into the scanner bore. Scan parameters were: 0.9mm isotropic resolution, TR/TE = 1000/121 ms, flip angle = 115-140°, FOV = 320x320x52mm. The regions of the phantom covered by both the upper and lower scans were intensity thresholded to segment out the water-filled chambers in the phantom. A gradient non- linearity unwarping algorithm [3] was then used to correct the spatial distortions for both the upper and lower scans, using vendor-supplied spherical harmonic coefficients as the initial input variables. The unwarping of this algorithm was then numerically optimized (BFGS method) for the coefficients corresponding to the Z axis to identify coefficients maximizing the dice coefficient of the segmented phantom chambers in the upper and lower scans (Fig. 2). After gradient non-linearity unwarping using either the vendor-supplied (default, Fig. 3A) or the optimized coefficients (Fig. 3B), PropSeg (Spinal Cord Toolbox) was used to determine the average SCCSA of a 4.5mm [≈ length of average red ant] region of the spinal cord at each vertebral level between C2 and T10 for upper and lower scans separately [4]. Appreciable differences in SCCSA between the default and optimized unwarping can be observed for both upper and lower scans in all four subjects (Fig. 4), especially towards the edge of the FOV (i.e. Z gradient). An illustrative example (Fig. 4C) was shown in Fig. 5 with SCCSA differences between the default and the optimized unwarping.

121

Conclusion By using the self-consistency criterion of the overlapping region of the in situ grid phantom, we were able to numerically optimize the spherical harmonic coefficients for the Z gradient. This potentially produces more accurate gradient non-linearity unwarping, which translates to more accurate SCCSA measurements. This in situ phantom and gradient non-linearity optimization algorithm may be used to provide scanner-specific spherical harmonic coefficients, providing improved reproducibility of SCCSA measurements between sites and vendors.

Clinical Relevance Spinal cord cross sectional area (SCCSA) holds promise as a biomarker of neurological disorders. However, the large FOVs required to obtain SCCSA from a large portions of the spinal cord are accompanied by significant spatial distortions due to gradient nonlinearity. While MRI vendors supply spatial unwarping algorithms, site- specific variations in the gradient linearity are present, which affects the reproducibility of longitudinal and multi-site studies.

References [1] Freund P et al, MRI investigation of the sensorimotor cortex and the corticospinal tract after acute spinal cord injury: a prospective longitudinal study, Lancet Neurol. 2013 Sep;12(9):873-81 [2] Liu, W. et al. In vivo imaging of spinal cord atrophy in neuroinflammatory diseases. Ann. Neurol. 76, 370–378 (2014). [3] Jovicich J et al, Reliability in multi-site structural MRI studies: Effects of gradient non-linearity correction on phantom and human data, Neuroimage. 2006 Apr 1;30(2):436-43 [4] De Leener B et al, Robust, accurate and fast automatic segmentation of the spinal cord, Neuroimage. 2014 Sep;98:528-36

Figures and tables Fig. 2 Spatial Fig. 1 correction of Fig. 3 Design and the grid phantom Unwarping structure using the results for a of the grid vendor-supplied representative phantom (A) unwarping subject using in sagittal coefficients (A) both the vendor (B) and and numerically supplied coronal (C) optimized unwarping views on coefficients coefficients CT, with a (B). The phantom (A) and sagittal in blue is from numerically line graph the upper scan, optimized (D) of the red is from the coefficients boundaries lower scan and (B) with upper in the CT the yellow and lower image. regions are images from two where both table positions intersect. merged to highlight mismatching areas in A.

Fig. 5 SCCSA measurements from a single subject (A, same as Fig. 4C) and the difference in SCCSA between the segmentations from the default and optimized unwarping algorithm (B).

Fig. 4 SCCSA measurements in all four subjects with both unwarping methods. 122

Modular toolsets for neurosurgery simulation and visualization

Authors & Affiliations Anthony B Costa, Joshua B Bederson Department of Neurosurgery, Icahn School of Medicine at Mount Sinai

Introduction The practice of pre- and intra-operative interactive visualization and modeling continues to grow as its value to clinical practice is augmented by new technologies, such as virtual and augmented reality, or 3D printing. Current tools which extract the necessary structural information from medical imaging modalities and allow virtual or other interrogation of the data are either difficult to use in a practical clinical setting, or sufficiently simple as to limit the knowledge available to the operator. Nonetheless, the broader medical visualization and simulation communities have invented tools which enable automated segmentation and interrogation of structures critical to the success of surgery, such as cranial nerves, vasculature, and cortical and subcortical parcellations.

Methods & Results We leverage these tools as inputs to a novel collection of simple to use utilities for neurosurgery simulation. Our pipeline is compatible with ATLAS-based subcortical volumetric segmentation (e.g., Freesurfer, ANTS), or any structural input in mesh- or voxel-based formats, together with volumetric data. The visualizer, based on VTK7’s OpenGL3x rendering backend, is efficient enough to display an arbitrary number of input structures or volumes at interactive refresh rates. Structures can be manipulated by adjusting parameters for each structure independently (e.g., color, opacity). Standard ATLAS-based and ITK/VTK-based tools are included in the pipeline directly. Also included is a novel volumetric shift-based segmentation tool, allowing an operating scientist to easily include information detailing aberrant pathologies rapidly and with minimal semantic information.

Conclusion We demonstrate these tools for a variety of cases, including tumor, vascular, hemorrhagic stroke, and spine. Its performance sufficient to run and be used on a laptop computer and capabilities for pre-operative planning through 3D printing the generated structures.

Clinical Relevance We find that repurposing the power of existing segmentation tools within a novel modular, multi-modal framework enables robust neurosurgical simulation for pre- and intra-operative planning with features not possible in any existing integrated simulation platform.

123 124

Connectomics signature of disease expression and risk to bipolar disorder

Authors & Affiliations G. Doucet1, D. Glahn2, S. Frangou1 1: Icahn School of Medicine at Mount Sinai, New York, NY. 2: Yale University School of Medicine, New Haven, CT.

Introduction There is growing evidence of aberrant resting-state functional connectivity (rs-FC) in patients with bipolar I disorder (BD). First-degree relatives of patients are at high risk of BD and this may be mediated through abnormalities in rs-FC. Therefore, in this context, we tested the power of graph theory metrics to distinguish between BD patients, their unaffected siblings and healthy volunteers.

Methods & Results We collected fMRI resting-state data from 78 patients with DSM-IV bipolar I disorder, 64 of their unaffected siblings (SIB), and 41 matched healthy volunteers (HV). After standard preprocessing of the data using SPM12, the brain was parcellated into 638 regions, and correlations were individually computed between each pair of regions’ time-series. The correlation matrices were then used to construct binary undirected graphs. In order to address the issue of disconnectedness, we built individual graphs in the range 1-100% in increments of 1%, based on the minimum spanning tree. Network topology was studied using graph theory implemented in the brain connectivity toolbox. We extracted the following metrics: global efficiency, characteristic path length, global clustering, small-worldness. We also estimated local metrics: clustering coefficient, local efficiency (Eloc) and participation. To assess the between-group differences in the measures of functional network organization, we used the Kruskal-Wallis test. Results: No significant group differences were observed for any of the global metrics tested (p<0.05, FDR corrected; Fig. 1). Group differences were only found at the local level. We found that BD patients had less connections in multiple regions such as in the primary cortices, the superior parietal and supramarginal gyri (Fig. 2). Importantly, a negative association was found between the BPRS (Brief Psychiatric Rating Scale) and Eloc in the middle occipital gyrus for the BD patients. The SIBs also showed less connection in these regions than in the HVs. Lastly, when comparing the SIB and the BD groups, we found reduced Eloc in the middle occipital gyrus and increased Eloc in the dorsolateral prefrontal cortex, the superior parietal gyrus, caudate nucleus and the rectus for the BDs, relative to the SIBs.

Conclusion The measures reflecting the global brain organization did not differ between our experimental groups, suggesting that BD, when in remission, may not affect the global network interconnectivity and therefore the modular brain organization. However, at the local level, graph-theory analysis showed brain topology differences reflecting disease expression in patients and vulnerability in their siblings.

Clinical Relevance These analyses provided a unique window into both disease-specific and disease-vulnerability brain functional reorganization.

125

Figures and tables

Figure 1: Global topological measures of the brain in Bipolar (blue), siblings (red) and healthy controls (green). A: Average Clustering; B: Characteristic Path Length; C: Global Efficiency; D: Small-Worldness.

Figure 2: Regions showing different local efficiency (Eloc) between groups. (A) shows the regions with different Eloc between the BD and HV. (B) shows the regions with different Eloc between the SIB and HV. (C) shows the regions with different Eloc between the BD and SIB groups.

126

Functional network organization in bipolar I disorder

Authors & Affiliations G. Doucet1, D. Glahn2, S. Frangou1 1: Icahn School of Medicine at Mount Sinai, New York, NY. 2: Yale University School of Medicine, New Haven, CT.

Introduction There is growing evidence of aberrant resting-state functional connectivity (rs-FC) in patients with bipolar I disorder (BD). First-degree relatives of patients are at high risk of BD and this may be mediated through abnormalities in rs-FC, and especially in their functional brain organization (e.g. network definition and connections). Therefore, in this context, we investigated whether BD patients, their unaffected siblings and healthy volunteers differ in the properties defining their brain’s functional network organization.

Methods & Results We collected fMRI resting-state data from 78 patients with DSM-IV bipolar I disorder, 64 of their unaffected siblings (SIB), and 41 matched healthy volunteers (HV). The individual data were preprocessed using SPM12. For each individual, we revealed brain functional organization by identifying groups of brain regions that were densely interconnected by strong rs-FC, using modularity maximization. To do so, we utilized the Louvain-like locally greedy algorithm implemented in Matlab (gounlouvain.m). It has been shown that increasing the resolution parameter ɣ yields refined modules. While a ɣ value of 1 is the most common choice in the literature, we chose to gradually increase the resolution parameter ɣ from 0.5 to 1.5 (by 0.1 gap) to test the reliability of the community structures (e.g. brain functional partition into networks). For each individual, we extracted the number of networks (also called modules) and the size of the largest network. Lastly, we computed an index of spatial similarity between each individual partition and the average normative partition (based on our control sample) (Z-score of Rand Coefficient). Results: We estimated the functional network organization for each individual. We did not find any significant differences in the number of networks (also called modules) estimated, the size of the largest network or the spatial similarity of the partitions between the groups (p>0.05), regardless of the gamma tested (Figure 1).

Conclusion These analyses suggest that BD does not affect the functional organization of the brain networks.

Clinical Relevance This analysis helps us to understand the effect of Bipolar I disorder on the brain functional networks at rest.

127

Figures and tables

Figure 1: Properties of the brain functional network organization in Bipolar (blue), siblings (red) and healthy controls (green). A: Number of Networks; B: Size of the largest Network; C: Z-score of Rand Coefficient (reflecting Spatial Similarity).

128

B1-Insensitive Simultaneous Multi-Slice DWI at 7T using SEAMS PINS

Rebecca Emily Feldman1, Rafael O'Halloran1, and Priti Balchandani1 1Translational and Molecular Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Introduction: The higher signal-to-noise-ratio (SNR) offered at ultra-high magnetic field strengths, such as 7 Tesla (7T), has been shown to increase the resolution of diffusion MRI (dMRI) as well as the precision and directional certainty of diffusion- based parameters [1,2] However, two major drawbacks of 7T dMRI include lengthy acquisition times and signal loss due to increased inhomogeneity in the applied radiofrequency (RF) field (B ). Simultaneous multi-slice (SMS) imaging using multiband excitation and parallel imaging is becoming more common for accelerating full brain acquisitions [3]. While SMS reduces the duration of the acquisition, the refocusing pulses typically used in dMRI are particularly sensitive to B non-uniformities leading to a loss in signal, or even complete signal dropout in parts of the image. SEmi-Adiabatic Matched-phase Spin-echo (SEAMS) Power Independent of the Number of Slices (PINS) is an SMS technique that can accelerate imaging while reducing a spin-echo based sequence's sensitivity to B non-uniformities [4] . We have created a dMRI sequence with SEAMS PINS excitation and refocusing and an EPI readout that provides improved immunity to B -inhomogeneity with RF power deposition, as measured by the specific absorption rate (SAR), comparable to existing sequences.

Methods: We created a SEAMS PINS pulse-pair with an excitation pulse duration of 21 ms and a refocusing pulse duration of 14 ms (Figure 1). The pulse-pair was created to match a refocusing pulse with a bandwidth of 1.11 kHz sampled by 94 lobes, creating a slice separation/slice thickness ratio of 12. The pulse-pair was inserted into a dMRI sequence with an EPI readout (Figure 2), and EPI navigators acquired pre-scan [2]. Diffusion-weighted images were acquired using 2 diffusion sequences 1) the SEAMS PINS pulse-pair and 2) a standard non-adiabatic single-refocused SMS diffusion pulse-pair with a time-shifted RF pulse multiband factor of 2 [5]. Imaging was performed in a spherical water phantom on a 7T whole body MRI system (Siemens, Erlangen). The acquisition was made with: b = 1500 s/mm ; TE = 73 ms; FOV = 21 cm x 21 cm; matrix = 210 x 210; Time = 9 minutes. Each sequence was acquired at four power settings, at adiabatic threshold or with the power calibrated to produce the expected flip angle at the center of the magnet (P ), at 1.2 x P , 1.4 x P , and 1.6 x P .

Discussion/Results: The SAR ratio for each sequence was noted, although none of the sequences exceeded the SAR limits for the scanner. The SEAMS PINS diffusion pulse deposited 80% of the power of the time-shifted sequence, for the same TR and scan times. The B -insensitivity of the sequence is shown in Figure 3. For a central slice of the phantom (Figure 3A), b = 1500 s/mm , the signal received on a projection through the phantom (Fig. 3A white line) is plotted for each power setting for both sequences. The signal produced by the non-adiabatic SMS diffusion sequence shows high sensitivity to B across the initial projection through the phantom. As the transmit B increases by 20%, most of the signal in the center of the phantom is destroyed, and as the B continues to increase, a larger region of the phantom experiences complete dropout. In contrast to this, the signal produced by the SEAMS PINS pulse-pair, is more consistent across the body of the phantom. It shows much less variation over the phantom cross-section as the power is increased by 20%, and even at 1.6 x adiabatic threshold, no portion of the phantom experiences complete signal dropout. At ultra-high fields, dMRI using SEAMS PINS can achieve accelerated simultaneous multi-slice imaging while restoring diffusion-weighted signal to regions that would experience excessive signal dropout using non-adiabatic pulse-pairs. The sequence is substantially more robust to B inhomogeneity, while maintaining SAR deposition levels similar to those of existing non-adiabatic sequences. Future work involves further reduction of minimum achievable echo time and validation of the sequence in vivo.

Acknowledgements: NIH-NINDS R00 NS070821. Thanks to Dr. Junqian (Gordon) Xu for helpful discussions.

References: [1] Polders DL et al 2011 MRM 33:1456-63 [2] Dyvorne et al MRM (epub) June 2015 [3] Eichner et al 2014 MRM 72:949-58 [4] Feldman et al MRM (epub) March 2015 [5] Auerbach et al 2013 69:1261-7

129

Figures

Figure 2: SEAMS PINS Matched-Phase Pulse-pair and Slice Select Figure 1: Diffusion pulse sequence including the pre- Gradients scan acquisition of 3 navigator lines

Figure 1: A) Projection through water phantom. B) Signal response to transmiter at 1x, 1.2x, 1.4x and 1.6x the power threshold. C) Signal response in time-shifted diffusion pulse to transmit at 1x, 1.2x, 1.4x, 1.6x power threshold.

130

Diffusion Method to Image Normal Human Optic Nerve

Lazar Fleysher1, Matilde Inglese2, Mark J Kupersmith3, Niels Oesingmann4

1. Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States. 2. Dept of Neurology, Radiology and Neuroscience, Icahn School of Medicine at Mount Sinai, New York, NY 10029 United States. 3. Professor of Neurology and Ophthalmology, Icahn School of Medicine at Mount Sinai, New York, NY 10029. 4. Siemens Healthcare, USA

Introduction A critical amount of functioning axons of the optic nerve is required to provide vision, pupillary function, motor and circadian rhythm control and other functions. The optic nerve is subject to a wide range of pathology including congenital anomalies, ischemic, inflammatory, neurodegenerative, metabolic and neoplastic diseases. Unfortunately, routine ophthalmologic examination usually can only reveal abnormalities of the fundus oculi and optic nerve head. MRI of the human optic nerve remains a challenge [1-4] due to its size, movement and surrounding CSF. The goal of this study is to explore the feasibility of routine diffusion-based MR imaging of the optic nerve using vendor-provided software and hardware.

Methods & Results Three healthy volunteers (one female) were studied. Images were acquired with a 3T Skyra (Siemens, Erlangen, Germany) equipped with the vendor-provided 32 channel head coil and with 45mT/m imaging gradients. Diffusion-encoded data were acquired using standard 2D single-shot DTI-EPI with 21 contiguous 2mm thick slices and 164mm x 48.2mm FOV with 102 x 30 imaging matrix. The short EPI echo train allowed significant reduction of image distortions associated with the magnetic field inhomogeneities. The slice package was positioned perpendicular to the optic nerve posterior to the globe. Diffusion-encoding direction was set along the slice-encode direction. During the scan, the subjects were instructed to fixate on a cross presented on the stimulus display. Other sequence parameters were TR=6000ms TE=84ms and 15 b-values ranging from 0 to 1000s/mm2 with 4 averages each. Fat suppression was “on”, CSF suppression was “off”. To avoid the wrap- around artefact in phase-encode direction (L/R), outer volume suppression was employed. The apparent diffusion coefficient along the nerve was computed on 4 slices posterior to the globe with clear spatial separation between CSF and the nerve. The obtained values (mean +/-std) for the optical nerve was 1630+/- 193mm2/s and for the CSF 2790+/-103mm2/s. These values correspond to those obtained previously using specially-developed MRI sequences [1-4].

Conclusion In this work we demonstrate an application of a single-shot EPI diffusion sequence with outer-volume suppression to optic nerve imaging. The advantage of this approach is that it is simple and is available on clinical systems. This paves a way to a routine diffusion-encoded clinical examinations of the optic nerve.

Clinical Relevance MRI of the human optic nerve remains a challenge due to its size and movement and surrounding CSF, despite major advancements in MR application. Yet MRI is needed to study the vast number of disorders that injure the optic nerve and/or reduce retinal processing and the effects of neuroprotection and restoration. We demonstrate that routine diffusion-based MRI examination of the optic nerve is possible using vendor-provided hardware and sequences (i.e. single-shot EPI sequence with outer-volume suppression).

131

References: 1. Wheeler-Kingshott et.al. MRM 56:446(2006) 2. Trip et.al. NeuroImage 30:498(2006) 3. Bodanapally et.al. AJNR 36:1536(2015) 4. Xu et.al. NMR Biomed 21:928 (2008)

Figures and tables

Diffusion weighted signal as a function of Twenty-one contiguous diffusion-encoding. diffusion slices overlayed on the sagittal and axial localizers.

132

Identification of brain regions implicated in auditory hallucinations as potential target sites for neuromodulation

Authors & Affiliations Nigel I. Kennedy, Won H. Lee, Sophia Frangou Department of Psychiatry, Icahn School of Medicine at Mount Sinai, 1. Gustave Levy Place, New York, NY, 10029

Introduction Auditory hallucinations are a common symptom of psychotic illness which can often remain refractory to treatment despite adequate trials of antipsychotic medications. The pathophysiology of this phenomenon is not well understood and a number of imaging studies have been published using different modalities to attempt to identify functional and neuroanatomical regions correlates. A number of coordinate based meta analyses of each data type have been published. By combining the data from these previously published studies it should be possible to more accurately identify the brain regions most commonly implicated in the pathophysiology of the condition. Furthermore, identification of these important brain areas has the potential to highlight target regions for use in neuromodualtion therapies.

Methods & Results

A literature search of PubMed was conducted using the terms “MRI”, “Hallucination”, “PET”, and “SPECT”. 6 meta analyses were identified, covering a total of 39 individual studies, data from 1,615 patients in total. These were separated into three groups: resting state studies, in which subjects with known hallucinations were studied at rest and typically asked to indicate when symptoms were occurring, functional studies in which patients with known hallucinations were asked to perform a task which recruited the auditory system, and morphometric studies which examined brain volumes in patients with known hallucinations. An additional search for data not included in these meta-analysis was also conducted which did not yield any additional results.

The Talairach coordinates from brain regions most strongly correlated with auditory hallucinations in each meta- analysis were compiled according to study modality. The coordinates highlighted as most important in the results of each meta-analysis were collected and separated and mapped to whole brain regions, color coded according to study modality. The resulting montage of these results is shown in Figures 1 and 2.

From these results and the visual depictions in the figures that there exists a concentration of implicated brain regions in the temporo-frontal region which may be important in the underlying pathogenesis of auditory hallucinations. Given the large number of individual patients included in this review and the identification of common neuroanatomical regions in studies using different methodologies, this area would appear to have the most potential as a target region for neuromodulatory intervention.

Conclusion Data compiled from meta-analysis of resting state, functional and morphometric MRI imaging studies of auditory hallucinations show a convergence of commonly implicated brain regions including the superior temporal gyrus, the insula and the post central gyrus. This study compiles results from a number of meta analyses and the convergence of these results on this neuroanatomical network would implicate these regions as important in the pathogenesis of auditory hallucinations.

133

Clinical Relevance

The finding of similar brain regions implicated by different study modalities in auditory hallucinations would suggest an important role for the underlying neurocircuitry in the pathogenesis of these symptoms. These findings highlight these brain areas as potential target sites for neuromodulation. It is possible that these areas could be used as the basis for interventional methods such as transcranial direct current stimulation (tDCS) or transcranial magnetic stimulation (TMS) which may be used as a treatment for this significant and often debilitating condition.

Figures and tables

Figure 1. Neuroanatomical regions most strongly implicated in auditory hallucinations from meta-analysis of resting state (red), task functional (green) and morphometric (blue) studies.

134

Spinal Cord Segmentation and Cross-sectional Area Calculation

Authors & Affiliations Joo-won Kim, Junqian Xu Icahn School of Medicine at Mount Sinai

Introduction Quantifying cross-sectional areas of the spinal cord accurately is important for monitoring disease progression (i.e., spinal cord atrophy) in neurodegenerative diseases, such as multiple sclerosis or spinal cord injury. Recently, the Spinal Cord Toolbox, developed by Polytechnique Montréal, McGill University, and Aix- Marseille Université, has provided a convenient tool for automated spinal cord segmentation and cross-sectional area quantification. In this study, we estimated a spinal cord inner line and fed it to the Spinal Cord Toolbox segmentation algorithm (i.e., PropSeg) to improve the spinal cord segmentation in non-optimal signal-to-noise ratio (SNR) and/or contrast-to-noise ration (CNR) images. In addition, we introduce a robust method to approximate the tangent vector to the cross section of the spinal cord.

Methods & Results Methods: The inner line of a spinal cord in an MR image was extracted by finding the shortest path between two voxels on a graph with proper edge weight (Fig. 1A,B). The segmentation results from the PropSeg with and without the inner line were evaluated. Instead of using a single estimated tangent vector for calculating a cross-sectional area, multiple (2,881) vectors were generated to estimate the tangent vector more accurately (Fig. 1C). The cross-sectional areas of the spinal cord estimated from a single vector and multiple vectors were assessed. Results: Among five healthy adult subjects, the inner line extraction improved the segmentation results in two of the cases that the default PropSeg did not segment properly. For the other three subjects, the PropSeg with or without inner lines performed comparably well (DICE coefficients > 0.9, Fig. 2A-D). The CUDA (version 7.5) implementation (Python Numba) was 3X faster than the CPU implementation (Python). The cross-sectional areas of the spinal cord estimated from multiple normal vectors were up-to 9% smaller than those from a single normal vector (Fig. 2E). The CUDA implementation was 10X faster than the CPU implementation.

Conclusion The inner line extraction dramatically improved the segmentation result in low SNR/CNR images. The multiple normal vectors moderately improved the accuracy of cross-sectional area calculation, considering that the expected area difference between two normal vectors is 1.5% when the angle between the vectors is 10 (cos(10) = 0.985). The CUDA implementation of both inner line extraction and cross-sectional area calculation greatly reduced the execution time.

Clinical Relevance

135

Figures and tables

Figure 1. Schematic diagrams of the proposed inner line extraction and varying normal vectors methods. In (A), the red curve is the inner line, calculated from the shortest path between the two cyan color end voxels. In (B), the node Vi has 26 edges and 8 of them are shown. The Euclidean distance between Vi and Vj is square root of 2. In (C), two tangent vector estimation (black and green) are shown on a spinal cord center line (red).

Figure 2. (A-D) Segmentation results of two subjects (A/B and C/D) using default PropSeg options (A, C) or inner lines (B, D). (E) (A) The box plots of relative differences of the cross-sectional areas between a single normal vector and multiple normal vectors at all center line points (~ 200-400 points).

136

Quantitative evaluation of simulated functional brain networks in graph theoretical analysis

Authors & Affiliations Won Hee Lee, Sophia Frangou Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA

Introduction Graph theoretical approaches to resting-state fMRI have been widely used to quantitatively characterize functional network organization in the resting brain, but mechanistic explanations for how resting-state brain works are still lacking. Whole-brain computational models have shown promise in enriching our understanding of mechanisms contributing to the formation and dissolution of resting-state functional patterns. It is therefore important to determine the degree to which computational models reproduce the topological features of empirical functional brain networks. Here, we focus on the performance of the Kuramoto model as it is considered most representative model of coupled phase oscillators and is widely used in the literature.

Methods & Results Empirical and simulated functional networks were defined based on 66 brain anatomical regions (nodes). Simulated resting-state functional connectivity (FC) was generated using the Kuramoto model constrained by empirical structural connectivity and the Balloon-Windkessel model. We applied graph theoretical approaches to optimally simulated FC and empirical FC data to characterize key topological features of brain networks. Finally, we quantified and compared the difference, in terms of relative error, in graph theoretical measures between the simulated and empirical functional networks. The averaged relative differences in graph theoretical measures were found to be 2–77% over the entire range of connection densities as well as 0.1–22% over a range of 37–50% connection densities.

Conclusion Our findings suggest that simulated functional data can be used with confidence to model graph measures of global and local efficiency, characteristic path length, eigenvector centrality, and resilience to targeted attack and random failure. Our results also highlight the critical dependence of the solutions obtained in simulated data on the specified connection density.

Clinical Relevance This study demonstrates the value of computational models in assessing whole-brain network connectivity, and provides a method for the quantitative evaluation and external validation of graph theory metrics derived from simulated data that can be used to inform future clinical study designs.

137 138

Simulating the effects of structural disconnection on brain dynamics in a large-scale model of auditory hallucinations

Authors & Affiliations Won Hee Lee, Sophia Frangou Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA

Introduction Auditory hallucinations (AH) are a frequent symptom of schizophrenia that occurs in about 50-70% of patients, but in 25-30% of patients, these symptoms remain refractory to drug treatment. The availability of whole-brain maps of anatomical connections together with computational models of the large-scale neural dynamics have shed light on the relationship between anatomical and functional connectivity. Importantly, they can be used to predict the effects of structural alterations (e.g., lesions) on brain dynamics for understanding disease states resulting from altered structural connectivity. Here, we focus on the effects of structural disconnections on the neural dynamics in a large-scale AH model.

Methods & Results We determined the nodes corresponding to anatomical regions implicated in AH, including caudal anterior cingulate cortex, pars opercularis, pars triangularis, superior temporal cortex, temporal pole, and trasverse temporal cortex. Structural networks in these regions were damaged by deletion of links on nodes in descending order of their degree. We incrementally removed structural connection of the AH nodes from the original structural matrix. To progressively simulate a disconnection in AH association areas, we constructed a sequence of pruned matrices (12 disconnected matrices) from the original network. We simulated brain dynamics using the Kuramoto model coupled with each of the pruned matrices and the intact matrix. We evaluated synchrony and metastability both globally and in AH association nodes. The percentage changes in global synchrony and metastability were found to be -20–20% and -28–24%, respectively, while the values for local synchrony and metastability were respectively -56–0.1% and -25–12% for a range of global coupling strength.

Conclusion This study demonstrates the capability of whole-head computational models to provide quantitative insight to understand the impact of disrupted structural networks on neural dynamics at either a global or a local level. The performance of simulated brain dynamics needs to be evaluated by comparing to empirical data in future studies.

Clinical Relevance Whole-head computational models may help the interpretation of clinical AH studies and explore the underlying mechanisms of anatomical network abnormaliteis in schizophrenia for improved efficacy in the treatment of the symptoms of schizophrenia.

139

Figures and tables

Figure 1. Percentage change in global and local measures of synchrony (left) and metastability (right) for a range of global coupling strength with different pruned matrices ranging from 0.6% to 4%.

140

Trajectory of general anesthetic action delineated by multi-echo multi-band functional MRI

Joshua S. Mincer1 and Prantik Kundu2,3 1Department of Anesthesiology, 2Department of Radiology, 3Department of Psychiatry

Introduction Although general anesthetic drugs are administered daily to tens of thousands of people in the United States, their mechanism of action remains enigmatic. The study of general anesthesia is complicated by its being a true multiscale problem, ranging from the molecular pharmacology of anesthetics to their ultimate effects at the whole brain (indeed, whole body) level. A further complicating factor is the difficulty (both practically and theoretically) of separating the effects of anesthesia from those of the surgery for which such anesthesia is administered, given that surgery itself is a complex stressor with effects on the brain and body that remain controversial. To overcome this limitation, we are studying healthy human volunteers awake and under general anesthesia using functional MRI, in the absence of surgical stimulation, at multiple timepoints in order to delineate a trajectory of anesthetic action in the peri- and post-anesthetic periods.

Methods & Results Resting state scans were obtained for 10 human volunteers with a Siemens Magnetom 3T MRI with 32-channel transmit-receive head-coil (Siemens, Erlangen, Germany) implementing a multi-band multi- echo fMRI pulse sequence (whole-brain 3mm iso.; TR=1.85s, 3 TEs; MB=2; GRAPPA=2) and analyzed with multi- echo independent components analysis (ME-ICA; Kundu et al. 2011). Scans were obtained at the following timepoints: prior to induction, at multiple points at depth of 1 MAC (minimum alveolar concentration) of sevoflurane anesthesia, shortly after emergence, and at one day and again at one week later. BOLD functional networks were automatically identified and differentiated from non-BOLD artifacts based on their T2* decay for all scans. The number of BOLD components identified defined the dimensionality. An increase in dimensionality may correspond roughly with increased complexity in signal at the local voxel level or decreased connectivity at the global level. Data demonstrate two distinct trends in the dimensionality trajectory in patients under anesthesia. In roughly half of the subjects, dimensionality decreased under anesthesia as compared to pre-anesthesia. In the others, dimensionality increased under anesthesia. In both cases, dimensionality returned to baseline one day later (if not shortly after emergence) and generally remained the same one week later.

Conclusion By leveraging advanced functional MRI methods, this research is elucidating the neural correlates of general anesthetic action in the human brain and the trajectory of anesthetic effects in the peri-anesthetic period and beyond, differentiated from effects of concomitant surgical stimulation. Preliminary dimensionality analysis suggests that there exists more than one such trajectory in that some subjects evince an increase in dimensionality under anesthesia as compared to the awake state whereas others demonstrate the opposite. In both cases, dimensionality returns to baseline (pre-anesthesia) at one day post-anesthesia and remains essentially unchanged one week later. Future timeseries analysis of the ME-ICA processed fMRI signals will uncover the specific neural dynamics (at the anatomic level) underlying the differences in anesthetic trajectory and develop a spatio-temporal map of general anesthetic action.

Clinical Relevance Identification of neural correlates of anesthetic action will enhance our understanding of the mechanisms of general anesthetic action (including loss of consciousness) at unprecedented spatio-temporal resolution in the human brain. Through delineation of the trajectory of anesthetic effects in the immediate peri- anesthetic period and beyond, we will ascertain the extent to which anesthetic exposure results in longer-term

141

chronic effects on neural dynamics. This has implications for the role (if any) of anesthesia in the clinical phenomenon of post-operative cognitive dysfunction (POCD), with specific relevance to the elderly population, as well as the issue of safety of anesthetic exposure, a question with specific relevance to the developing (pediatric) brain.

142

Increased Salience of Gains Versus Decreased Associative Learning Differentiate Bipolar Disorder From Schizophrenia During Incentive Decision Making

Dominik A. Moser1 & Sophia Frangou1 1Psychiatry, Icahn School of Medicine, Mount Sinai, New York, NY

Introduction Bipolar Disorder (BD) and Schizophrenia (SZ) are both associated with abnormalities in incentive decision making. We have previously shown that the underlying mechanism differs between the two disorders; reward prediction was disrupted in SZ whereas BD was associated with increased incentive salience of gains (Brambilla et al. 2013; doi: 10.1017/S0033291712001304). Here we focus on describing the neural correlates undelying these dissociable abnormalities in incentive decision making in BD and SZ.

Methods & Results We obtained 3T functional magnetic resonance imaging (fMRI) data from demographically matched patients diagnosed with BD (n=50) or SZ (n=50) and healthy controls (n=50) while performing an event related guessing task with monetary reward, developed by the Human Connectome Project (HCP). All patients were in remission. Whole brain volume analysis was conducted in SPM12 examined the effect of condition (reward, loss, neutral) and diagnostic group and their interaction.

Conclusion Compared to HC, patients with SZ showed (a) reduced dorsal and frontopolar activation regardless of condition, (b) reduced striatal activation during anticipation regardless of condition, and (c) normal striatal activation following reward or loss. Compared to HC, patients with BD showed (a) normal activation in prefrontal regions regardless of condition, (b) increased striatal activation during anticipation and following reward, and (c) normal striatal activation during anticipation and after loss. A diagnosis by group interaction was noted in the dorsal prefrontal and striatal regions.

Clinical Relevance Our findings support abnormal reinforcement learning in SZ and increased sensitivity to gains in BD.

143 144

Clustered, Connectivity-Based Surgical Planning for Deep Brain Stimulation

Authors & Affiliations Rafael O’Halloran1, Prantik Kundu1, Brian Kopell2 1ISMMS, Radiology, 2ISMMS, Neurosurgery

Introduction Deep brain stimulation (DBS) is used to treat a variety of neurological and psychiatric conditions including Parkinson's disease (PD). While DBS is often successful, it can fail to produce the desired effect as a result of sub- optimal target selection. The current model of targetting lacks a key element - axons of passage - which are highly sensitive to stimulation (McIntyre et al. 2004) and comprise a connectome that acts as a receptive field. Characterizing this connectome with white matter tractography is technically possible but visualization and interpretation of the data can be a challenge (Figure 1). Here we enable DBS targeting of structural networks based on clustering white matter tracts interconnecting targets, allowing simple overlay in the surgical planning software. We believe this approach may make connectivity-based targeting simple to implement and ultimately increase the efficacy of DBS.

Methods & Results Subjects: 2 Subjects with PD undergoing DBS of the caudal zona incerta were recruited and provided informed consent. MRI: MRI was performed prior to surgery and included a T1-weighted anatomical scan, diffusion- weighted imaging (1.8x1.8 mm in-plane resolution and 3 mm thick slices, 60 directions and 5 b0 scans) and quantitative susceptibility mapping. CT: CT was performed on the day of surgery with 0.6x0.6x1mm resolution. Post-operative CT was performed upon completion of the surgery. Surgery: DBS surgery was performed by an experienced neurosurgeon (Machado et al. 2006). Both patients received bilateral implants (Medtronic 3389) with each implanted on different days. Surgical planning was performed on a Steath Workstation (Medtronic, Minneapolis, MN, USA). Data Processing: Cortical and sub-cortical segmentation was performed from the T1- weighted image with Freesurfer. Fibre-tracking was performed using the MRtrix package (https://github.com/MRtrix3/mrtrix3) (Tournier et al., 2012). Connectivity matrices were computed for each voxel from the tractography seeded from that voxel and the freesurfer parcellation. Clustering: Clustering was performed in MATLAB using a k-means independently for the left and right sides. Images were generated by assigning a different color to each cluster. Figure 2 provides a visual guide to the workflow.

An example of the clustered connectivity in the ROI for subject 1 is presented in Figure 3. The T1-weighted image lacks contrast in the midbrain (Figure 3a). While the QSM depicts the STN and red nucleus (Figure 3b) it lacks connectivity information. Connectivity clusters show areas with similar connectivity (Figure 3c). Note that the clusters are grouped together spatially in a way that contours with anatomy. This is expected given that neighboring areas should have similar connectivity even though the k-means algorithm is agnostic to spatial location. The clustered connectivity in subject 2 is given in Figure 4 with 2 examples of average connectivity plots for 2 clusters - one nearest to the implanted electrode location on the right side (Figure 4b) and one just anterior to it (Figure 4c). Increased connectivity to the lateral and medial orbitofrontal cortex in the second ROI (Figure 4c) versus the first (figure 4b) is expected due to the limbic (more anterior) component of the STN's connectivity to the prefrontal cortex.

145

Conclusion A simple k-means-clustering-based method by which connectivity data can be used with standard surgical planing environments for DBS was presented. This technique may provide a useful way to quickly assess connectivity in the target area in presurgical planning. Ongoing work is focused at defining connectomic targets that can be used to identify target clusters. k-means was chosen for its simplicity, speed, and robustness, however more sophisticated algorithms will be explored such as principle components analysis. We believe these results establish the proof of concept and provide the foundation for future work.

Clinical Relevance: This technique if successful would allow fully automatic targeting for DBS. This has the benefit of providing not only more accurate targeting but also allowing surgery to be performed on a fully sedated patient since accurate targeting would obviate electrical recording in the awake patient.

Figures and tables

Figure 1: An example of fiber tracks in the whole brain (colored lines) overlaid on the T1-weighted pre-operative MRI (greyscale) Figure 2: The basic process and concept behind the and with the post-operative CT overlaid (translucent orange). clustering method.

Figure 4: a) Clustered connectivity overlaid on T1- weighted MRI and post-operative CT. b) The Figure 3: Images with the post-operative CT overlaid connectivity plot cluster closest to the electrode showing both electrodes. a) T1-weighted. b) QSM. c) location and, c) just anterior to the one in (b). Connectivity clusters show areas with similar contrast on the right side. 146

High-resolution intra-cortical myelin profiling at 7 Tesla

Emma Sprooten1, Rafael O’Halloran2, Won He Lee1, Gaelle Doucet1, Dominik Moser1, Alexander Rasgon1, Morgan Goodman1, Alejandro Paulino1, Sophia Frangou1

1Dapertment of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY 2Translational and Molecular Imaging Institute Translational and Molecular Imaging Institute and Brain Imaging Center, Icahn School of Medicine at Mount Sinai, New York, NY

Introduction Abnormalities in myelination have long been considered important in the pathophysiology of several neurological and psychiatric disorders. In vivo neuroimaging lends itself to the assessment of myelination, but progress to date has been hindered by methodological limitations, especially with respect to intracortical myelin, which is more directly related to neural function and plasticity than subcortical white matter. Here, we applied a previously validated method to extract intracortical myelin profiles the depth of the cortex in a high resolution anatomical scan.

Methods & Results MRI data was acquired on a 7T Siemens Magnetom scanner at the Translational and Molecular Imaging Institute at Mount Sinai. Three T1 maps were acquired using a Magnetization-prepared 2 rapid acquisition gradient echoes (MP2RAGE) sequence. The first covered the whole brain with an isotropic resolution of 0.7mm (TR=6000ms, TE=5ms, TI=1050ms). Two additional scans covered the ventral and dorsal parts of the brain separately, to allow for ultra-high 0.5mm isotropic resolution. Laminar myelin profiling was implemented using CBS tools (https://www.nitrc.org/projects/cbs-tools). The 0.5mm and 0.7mm scans were aligned. The cortical ribbon was segmented. Cortical depth was modeled (Figure 1A) and the cortex was divided into 20 cortical layers using a volume-preserving approach (Figure 1B; Waehnert et al., 2014). For each voxel in each cortical layer, T1 values were transformed by region-specific models based on cyto-architectural atlases (Dinse et al., 2015). Myelin values within each cortical layer were extracted for the motor cortex (BA4) and the somatosensory cortex (BA1). Modeled T1 values in BA1 and BA4 ranged between 1400 and 2400 (Figure 1D), where lower values reflect higher myelin concentrations. Values were lowest in de deepest layers of the cortex, as would be expected for a signal reflecting myelin concentration. These results are similar to a previously reported post-mortem validation of this method (Dinse et al., 2015).

Conclusion These preliminary results, of two well characterized regions within a single subject, support the feasibility and inter-site stability of intracortical laminar profiling in vivo.

Clinical Relevance The feasibility of this approach encourages its application to clinical studies, which will enable for the first time the quantification of intra-cortical myelin in patients in vivo without confounding of post-mortem tissue preparation. Importantly, this technique enables the longitudinal interrogation of this pathogenic mechanism in large samples of patients with neuropsychiatric and neurological disorders, and will facilitate the assessment and monitoring of new treatments aiming to improve myelin pathology.

147

Figure 1. A. Modeling of cortical depths using volume-preserving approach (Red=deep). B. Discrete sampling into layers using the volume-preserving approach. C. 3D rendering of cropped prefrontal cortex, with discrete layers visible along the rim. D. T1-values are transformed according to area-specific cyto-architectonic distributions. The resulting modeled T1-values are extracted from each layer within regions of interest. Note that higher T1-transformed values reflect less myelin. As expected and similar to previous post-mortem- validated results, the deepest cortical layers (V and VI) have the most myelin in BA1 and BA4.

A. C.

B.

D. Myelin profile 2500 BA1 BA4 2100

1700 modeled T1 value 1300 0 0.2 0.4 0.6 0.8 1 corcal depth

148

Functional integration of whole-brain networks during resting-state in cocaine addicted individuals

Authors & Affiliations Anna Zilverstand, Muhammad A. Parvaz, Scott J. Moeller, Rita Z. Goldstein Department of Psychiatry and Neuroscience, Icahn School of Medicine at Mount Sinai, New York

Introduction Previous studies in individuals with cocaine use disorder (iCUD) focused on describing changes in the functional connectivity of the brain’s reward system. Only a handful of studies applied whole-brain approaches, implicating other brain regions such as frontal cognitive control regions (Kelly et al., 2011), memory regions (Ding & Lee, 2013), and regions involved in self-referential processing (Konova et al., 2015). Here we investigate resting-state connectivity using measures of functional integration, employing complex network analysis to estimate the ease with which brain regions communicate with each other (Latora & Marchiori, 2001; Rubinov & Sporns, 2010). Two measures of functional integration were computed to compare individuals with non-recent [iCUD- (urine negative)] and recent drug use [iCUD+ (urine positive)]: local efficiency, summarizing functional integration within brain regions, and global efficiency, indicating the integration across whole-brain networks.

Methods & Results Ten minute resting-state functional magnetic resonance imaging (fMRI) scans were acquired in iCUD+ (N=26, age 47±8 yrs) and iCUD- (N=17, age 47±8 yrs) and healthy controls (N=32; age 40±8 yrs) matched on race and gender. Groups differed in age, depression scores (Beck Depression Inventory), smoking status and performance

IQ (Wechsler Adult Intelligence Scale, Matrix Reasoning); all used as covariates in the analysis. Participants’ drug use histories and baseline craving were assessed by a structured interview, the Cocaine Selective Severity Assessment and the Cocaine Craving Questionnaire. The drug use assessment revealed that iCUD+ demonstrated more chronic and severe drug use than iCUD-. This was evidenced by significantly increased lifetime use (+7 years, corrected for age), earlier age of onset (-3 years) and longer heavy use (+30%) with more frequent cocaine use (+10%) and more money spend (+159%). Further, iCUD+ reported more withdrawal symptoms (+73%) and higher subjective craving (+294%). The imaging data was preprocessed following standard procedures and parcellated using a high-resolution anatomical template. Local and global efficiency were computed based on each individual’s connectivity matrix. The results demonstrated a linear effect in decreased global efficiency in prefrontal cognitive control regions and subcortical brain regions including those implicated in reward processing, such as putamen (iCUD+iCUD->Controls) (figure 2).

Conclusion The results demonstrate whole-brain changes in functional integration going far beyond the reward network. We observed less integration of the frontal cognitive control network, reward and memory regions with the rest of the brain with more chronic use, indicative of less efficient recruitment of these cognitive networks. Further, functional integration within sensory networks and regions involved in value representation was increased.

Clinical Relevance The development of novel brain-based interventions may require targeting entire brain networks, rather than focusing on region-specific parameters of brain functioning. Results demonstrate that whole-brain resting-state connectivity could provide a tool to monitor disease progression and intervention in the addicted brain.

149 Figures Mean global efficiency in network with decreased functional integration

Controls iCUD- iCUD+ Figure 1. Global efficiency. Decreased global functional integration during resting-state was found in a network of frontal cognitive control regions (e.g., anterior cingulate, supplementary motor cortex), subcortical regions involved in reward processing/memory (e.g., putamen, hippocampus) and regions involved in language processing (e.g. Heschl’s/Wernicke’s), while increased global functional integration was found in regions involved in value representation (dorsolateral prefrontal; blue = decreased, red = increased, p < 0.05). Mean local efficiency in network with increased functional integration

Controls iCUD- iCUD+ Figure 2. Local efficiency. Increased local functional integration during resting-state was found in a network of frontal regions involved in value representation (e.g., dorsolateral prefrontal, inferior frontal, orbitofrontal), as well as in brain regions involved in sensory (e.g., visual, somatosensory) and multisensory processing (e.g., inferior/superior parietal; blue = decreased, red = increased, p < 0.05). References Ding, X., & Lee, S. W. (2013). Cocaine addiction related reproducible brain regions of abnormal default-mode network functional connectivity: A group ICA study with different model orders. Neuroscience Letters, 548, 110–114. doi:10.1016/j.neulet.2013.05.029 Kelly, C., Zuo, X.-N., Gotimer, K., Cox, C. L., Lynch, L., Brock, D., … Milham, M. P. (2011). Reduced interhemispheric resting state functional connectivity in cocaine addiction. Biological Psychiatry, 69(7), 684–692. doi:10.1016/j.biopsych.2010.11.022 Konova, A. B., Moeller, S. J., Tomasi, D., Volkow, N. D., & Goldstein, R. Z. (2013). Effects of Methylphenidate on Resting-State Functional Connectivity of the Mesocorticolimbic Dopamine Pathways in Cocaine Addiction. JAMA Psychiatry, 70(8), 857. doi:10.1001/jamapsychiatry.2013.1129 Latora, V., & Marchiori, M. (2001). Efficient Behavior of Small-World Networks. Physical Review Letters, 87(19):198701. Epub 2001 Oct 17. Rubinov, M., & Sporns, O. (2010). Complex network measures of brain connectivity: Uses and interpretations. NeuroImage, 52(3), 1059–1069. doi:10.1016/j.neuroimage.2009.10.003

150

Sponsors & Contributors

151 152

Sponsors

GOLD

153 Sponsors

Silver

154

Contributors

153 154 TMII Highlights 2015/2016

155

WHAT’S NEW FACULTY SPOTLIGHT SCIENCE SPOTLIGHT IMAGING SPOTLIGHT CORE SPOTLIGHT BIC CORNER Translational & Molecular Imaging Institute Fall, 2015 Issue 8 tmii.mssm.edu Message from the Director Hope you all had an exciting summer with This goes to all you as mentors, trainees, the ultra hight field 7T MRI work all featured some time for rest and rejuvenation. I write colleagues, friends and family. None of this in this Newsletter. Other announcement such this at the eve of a special day for me. As would have been possible without you. I am as the TMII lectures series, 6th Annual TMII featured in the issue, tomorrow I will receive also so proud of all the current TMII members Symposium (April 22, 2016), and the 3rd New an Endowed Professorship designated as the and their achievements, some of which are York Metro Imaging Research Consortium Mount Sinai Professor of Medical Imaging shared with you in this Newsletter. Your hard meeting (November 18, 2015). Again, I thank all and Bioengineering. I could not be happier work again, led to a most successful and of you for making all this possible and wish you with this honor and designation that features stimulating TMII Annual Symposium. We look a great TMII newsletter read. two passions I share in my career, Medical forward to the October 7 Brain Imaging Center Zahi Fayad, PhD Imaging and Engineering. However, the source Annual Symposium which promises to be as Director, Translational & Molecular of my biggest joy and gratitude comes from stimulating and successful. Your hard work Imaging Institute all the people who joined in the past and is also leading to top publications, patents, Professor of Radiology and Medicine join everyday in this wonderful adventure. research grants, and new application such as [email protected]

WHAT’S NEW? TMII News & Updates The 6th Annual TMII Symposium was another Calcagno - Mani and Hadien Dyvorne on their work on the development of efficient motion resounding success. There we 47 exceptional new faculty appointments to the department of compensation strategies for the delivery of abstracts submitted (up from 39 last year) and Radiology. quantitative and early diagnostic information 244 registered attendees - a 50% increase from from cardiovascular PET/MR imaging scans. TMII would also like to welcome two new post- last year! See the Imaging Spotlight feature for doctoral fellows. Alan Seifert, PhD comes to Lastly, TMII user Sophia Frangou, MD, PhD and more information on the meeting TMII by way of Dr. Felix Wehril’s lab at Univeristy colleagues recently publish a paper (HBM There are many congratulations to give out of Pennsylvania. At TMII, Dr. Seifert will be 36, 10, 4158-4163, Oct 2015) reporting on the this quarter; namely to TMII director Zahi Fayad studying the processing and transmission of interaction between personality and short- who is receiving the Medical Imaging and pain in the spinal cord and brainstem using fMRI term plasticity during working memory. The Bioengineering Endowed Chair on Oct 1 (see at 3T and 7T, and assisting with RF hardware- paper was picked up in July by the Smithsonian Faculty Spotlight feature for more details). TMII related projects. Nikolaos Karakatsanis, PhD magazine. would also like to congratulation Drs. Claudia has joined us from the University of Geneva to

UPCOMING EVENTS

> Thurs, Oct 1, 2015 1:15pm - 2:15pm - Hess Center, Seminar room B - Inaugural Medical Imaging and Bioengineering Lecture Jeff W.H. Bulte, MS, PhD - Professor, Johns Hopkins Medical - “Imaging cell delivery near the bed: Are we there yet?” > Wed Oct. 7, 2015 8:15am - 5:3pm - Hess Center, Davis Auditorium - BIC 2nd Annual Symposium Registration is open: https://bic.mssm.edu/blog/bicday/bicdayregistration/ > Fri, Oct 23, 2015 11am-12pm - Hess Center, TMII Large Conference Room s1-117 - TMII Lecture Series Charalampos (Harry) Tsoumpas, PhD - Lecturer, University of Leeds - “PET Imaging Reconstruction in a Nutshell” > Fri, Oct 30, 2015 11am-12pm - Hess Center, TMII Large Conference Room s1-117- TMII Lecture Series Irene Polycarpou, MSc, PhD - Lecturer, European University of Cyprus - “The importance of motion and attenuation correction in establishing PET/MR imaging as the new technique for early diagnosis and therapy monitoring” > Wed, Nov 18, 2015 10am - 2pm - Nathan S. Kline Institute for Psychiatric Research - 3rd New York Metro Imaging Research Consortium (NYMIRC) Abstract deadline October 18, 2015 - contact Tina Bermudez ([email protected]) or https://tmii.mssm.edu/nymirc/ for more details > April 22, 2016 8am - 5pm - 6th Annual TMII Symposium - Save the Date! More details to follow For more information on these and other events go to: http://tmii.mssm.edu/events

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | One Gustave L. Levy Place, Box 1234 | New York, NY 10029-6574 | tmii.mssm.edu FACULTY SPOTLIGHT Mount Sinai Professor in Medical Imaging and Bioengineering New Endowed Chair in Medical Imaging and Bioengineering Zahi A. Fayad, PhD Dr. Zahi Adel Fayad was born in Beirut atherosclerosis, with first demonstrations of in the American Heart Association. All this Lebanon and raised there and in France. He vivo plaque in transgenic mice and humans. They supplemented by several pharmaceutically had his training in Electrical and Biomedical were first to demonstrate that atherosclerosis funded multicenter clinical trials for the Engineering at Bradley University, the Johns of the aorta and carotid arteries can be clearly evaluation of novel cardiovascular drugs. Hopkins University and at the University of visualized, quantified Pennsylvania. In 1997 he joined the faculty and characterized by He received as Assistant Professor at Mount Sinai School magnetic resonance several honors and of Medicine where he was recruited by Drs. and that treatment with prestigious awards Valentin Fuster and Burton Drayer from the statins can influence its from the American Department of Radiology. progression. Heart Association, American College His laboratory has of Cardiology, significantly contributed Radiological Society to the field of vascular In vivo cross-sectional Black Blood-MR images of North America, positron emission of lumen (A) and wall (B) of proximal LAD from International tomography or PET normal subject (see arrow). Fayad et al. Circulation. Society of Magnetic imaging where he 2000;102:506-510 Resonance in recently showed the Medicine and results of the 1st noninvasive (MRI and FDG-PET) recently received the highest distinction multicenter clinical trial evaluating atherosclerosis award from his alma matter. using a new treatment. He is a caring mentor and teacher. He is Dr. Fayad now serves as Professor of Radiology In the field of molecular imaging, his group was the passionate about his work, his Institute and Medicine (Cardiology). He is the founding first to describe the use of a targeted iodine and and its members. Moreover, he is very Director of the Translational and Molecular gold based computed tomography nanoparticles passionate about high intensity exercise; Imaging Institute (TMII) and Vice chair for for imaging atherosclerosis. With his long-term sailing and enjoys running and racing in Research in the Department of collaborator Dr. Edward Central Park. He is married to Monique P. Radiology. Fisher who was initially Fayad, MBA and is the proud father of Chloé in the department of (13 year old) and Christophe (9 year old) and His interdisciplinary and Cardiology at Mount Sinai after spending seven years in Manhattan discipline bridging research - and now at NYU, and Dr. now lives in Larchmont. from engineering to biology Willem Mulder from Mount and from pre-clinical to Sinai they described the use clinical investigations - has of lipid multimodal (MR, CT, been dedicated to the optical, etc.) nanoparticles detection and prevention of for nanomedicine - cardiovascular disease with molecular imaging and drug many contributions in the field delivery. From this work, of multimodality biomedical 1st in man clinical studies imaging and nanomedicine. were recently conducted He, his team and collaborators for targeted nanotherapy of including Valentin Fuster, atherosclerosis. Dr. Fayad will be awarded the new helped pioneer, validate and endowed Chair in Medical Imaging and disseminate several novel He holds over 12 US and Bioengineering on Thursday October 1, imaging techniques and Worldwide patents and/or 2015 at 5pm in the Goldwurm Auditorium at strategies for the noninvasive Axial views of the same atherosclerotic patent applications. He has the Icahn School of Medicine. detection and the treatment plaque (white arrowheads) in the aorta of authored more than 300 of atherosclerosis. a rabbit, obtained by CT before (a), during well-cited publications in (b) and 2 h after the injection of N1177 (c) the field of cardiovascular Zahi Fayad, PhD His group introduced the use or a conventional contrast agent (d). Fayad medicine. He continues Director, Translational & Molecular Imaging Institute of noninvasive multimodal senior author (Nature Medicine 13, 636 - 641 to be extremely well (2007) Professor of Radiology and Medicine for the assessment of funded by the NIH and [email protected]

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | One Gustave L. Levy Place, Box 1234 | New York, NY 10029-6574 | tmii.mssm.edu 2 SCIENCE SPOTLIGHT Recent Discoveries Lead to New Patents

Doctors Zahi Fayad and Willem Mulder were the particles (or a solution of the material or recently awarded a patent (WO 2013192310 materials to form the particles and a non- A1) for “Mass production and size control solvent for the material or materials), at least of nanoparticles through controlled two symmetrical microvortices are formed microvortices”. Methods for making particles, simultaneously. The method can be used to such as nanoparticles, devices useful in the prepare polymeric or non-polymeric particles methods, and particles made by the method and hybrid particles, such as lipid-polymer are described herein. The methods involves hybrid particles, as well as such particles the use of a microfluidic device, such that containing one or more agents associated with upon mixing solutions of the materials to form the particles.

IMAGING SPOTLIGHT 5th Annual TMII Symposium - Highlights Bright Stars Reduction in Simultaneous Carotid PET/MR Lastly the best poster in the Cancer and Body Imaging.” Imaging section was award to Dr. Sean Carlin (MSKCC) for “Kinetic modeling of PET data In the Neuroimaging section Dr. Emma for the characterization of tumor perfusion Sprooten (Frangou Lab) was awarded for and hypoxia in response to VEGF signaling her poster “A comprehensive probabilistic blockade.” tractography study in sibling pairs discordant In addition to the outstanding keynote for bipolar disorder.” lecture, invited speakers and selected oral presentations, four posters stood out from The best poster in the Nanomedicine section the rest and were awarded top poster in was awarded to Dr. Constantinos Hadjipanaysis their session. In the Cardiovascular imaging (Neurosurgery) for “Radiosensitivity session, Mootaz Eldib (Fayad Lab) was enhancement of radioresistant glioblastoma given the award for his poster “Feasibility by epidermal growth factor receptor antibody- of 18F-Fluorodexyglucose Radiotracer Dose conjugated iron-oxide nanoparticles.”

CORE SPOTLIGHT Ultra High Field Imaging at 7 Tesla Siemens Magnetom 7T

This is an ultrahigh field 7.0 Tesla actively The (warm) inner bore of the magnet is 82 cm, and 32-channel Rx head coil and the 8-channel shielded whole body MRI scanner. The super- which houses the 60 CM inner patient bore. The Tx and 8-channel Rx head coil. conducting magnet is self-shielded, reducing dimensions of the magnet without covers is its overall footprint and making it compact and approximately 2.5 m in length, 2.6 m in width, lightweight by 7T standards, weighing 24-tons. and 2.65 m in height. The 5-Gauss line extends slightly further than for a 3T scanner with 5.6 m With less then 40 system across the globe, this radial and 7.8 m axial dimension. A whole-body system provides investigators the ability to get gradient system provides gradient amplitude unparalleled resolution and reductions in scan of up to 70 mT/m per axis, and a maximum time. slew rate of up to 200 T/m/s. The RF transmit system comes with 8 parallel transmit channels; 8 individually shaped RF pulses can be prescribed simultaneously and independently in amplitude and phase. The multi-nuclei package allows for imaging and spectroscopy at non-proton frequencies, i.e. detection of e.g. 19F, 31P, 7Li, 23Na, 13C, 17O. Our 7T/820AS is configured to accommodate an 8-channel Tx- Ultra high resolution T2 of a subject with array and 48-channel Rx receivers. Several coils epilepsy showing structural abnormalities in the hippocampus (inset). Courtesy of R. Feldman are currently available such as the 1- channel Tx

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | One Gustave L. Levy Place, Box 1234 | New York, NY 10029-6574 | tmii.mssm.edu 3 BIC CORNER There is still time to register for the Brain Moeller: K01 and R21; Bryan Denny: F32), clearly to substances (including nicotine, alcohol, Imaging Center’s Second Annual Symposium- demonstrating NIH’s substantial and marijuana), academic please visit https://bic.mssm.edu/blog/save- support for the neuroimaging achievement, cognitive skills, the-date-2nd-annual-bic-symposium/ to let work being developed by mental health, and brain us know you plan to attend. The symposium Mount Sinai’s BIC faculty. Of structure and function using program will include presentations on Technical particular note is a landmark advanced research methods. Innovations, Cognitive Interventions and study that has been launched BIC’s Rita Z. Goldstein will lead Connectivity and Multimodal neuroimaging by the NIH, the Adolescent this study at Sinai. with opening remarks by Drs Rita Goldstein, Brain Cognitive Development Zahi Fayad and Mount Sinai Health System CEO (ABCD) Study that will follow The neuroimaging facilities and President Kenneth Davis, and will include approximately 10,000 children available through TMII a keynote presentation by Dr. Nora Volkow, beginning at ages 9 to10, continue to develop - the Director of NIDA. Refreshments, lunch and a before they initiate drug use, installation of the behavioral wine and cheese reception will follow, so please through the period of highest stimulation and response let your colleagues know, and make plans to risk for substance use and recording apparatus to extend attend! other mental health disorders. functional MRI opportunities This study, encompassing a at the human 7 Tesla high- BIC offers congratulations to the users who have Coordinating Center and a Data field MRI scanner is nearing recently received funding from the NIH (Daniela Analysis and Informatics Center in addition to completion For other news, be sure to attend Schiller: R03 and R01; Emily Stern: R21/R33; Scott 11 Research Project Sites, will track exposure the BIC symposium next week (Oct 7)

CONTACTS Zahi A. Fayad, PhD Director, Translational and Molecular Imaging Institute Rafael O’Halloran, PhD Director, Cardiovascular Imaging Program Chief, Imaging Acquisition - BIC Faculty Professor of Radiology and Medicine (Cardiology) Assistant Professor of Radiology and Psychiatry [email protected] [email protected] Priti Balchandani, PhD Cheuk Y. Tang, PhD Director, High-Field MRI Program Director, Imaging Core Assistant Professor of Radiology and Neuroscience Associate Professor of Radiology and Psychiatry [email protected] [email protected] Prantik Kundu, PhD Bachir Taouli, MD Chief, Image Analysis Section & Advanced Functional Neuroimaging Section - BIC Faculty Director, Cancer and Body Imaging Program Professor of Radiology and Medicine Assistant Professor of Radiology and Psychiatry [email protected] [email protected] Venkatesh Mani, PhD Junqian Gordon Xu, PhD Cardiovascular Imaging Neuroimaging Assistant Professor of Radiology and Neuroscience Assistant Professor of Radiology [email protected] [email protected] Willem J. M. Mulder, PhD Christopher J. Cannistraci, MS Director, Nanomedicine Program Program Manager Associate Professor of Radiology Technical Operations Manager [email protected] [email protected]

Ways to keep in touch

Twitter: @TMIInyc Website: http://tmii.mssm.edu Facebook: TMII.SINAI Youtube: https://www.youtube.com/playlist?list=PLqLDR0CTP9_ Address: Leon and Norma Hess Center for Science and Medicine otAZpwEy3EgOStthPo7V9f 1470 Madison Avenue (between 101st and 102nd St) - 1st floor Linkedin: https://www.linkedin.com/groups/Translational- New York, NY 10029 Molecular-Imaging-Institute-TMII-8358896/about Numbers: Tel: (212) 824-8466 Fax: (646) 537-9589

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | One Gustave L. Levy Place, Box 1234 | New York, NY 10029-6574 | tmii.mssm.edu 4 WHAT’S NEW FACULTY SPOTLIGHT SCIENCE SPOTLIGHT IMAGING SPOTLIGHT CORE SPOTLIGHT BIC CORNER Translational & Molecular Imaging Institute Winter 2016 Issue 9 tmii.mssm.edu Message from the Director I write this as we are well into the brand new laboratory, showing the exiting work with our Finally, I need one big favor from you. Mount year after I hope for all of you some great 7T whole body MRI scanner in cervical spinal Sinai is putting together a new Strategic Plan December break celebrating the holidays cord and brainstem imaging. in advance of a capital campaign that will be with your loved ones and families. Indeed, launched in early 2017. TMII and Radiology here at TMII we continue to have reasons to We also are looking forward to celebrate two were charged with formulating a Strategic celebrate. One of our TMII family members, exiting upcoming events, April 8 the BIC 1st Plan (SP) that encompasses basic and clinical Dr. Willem Mulder has been promoted to Full annual 10k run, I know many of how have been imaging research and education. We have Professor with Tenure in the Department of training hard for this event and I look forward to already started the planning for the SP and we Radiology. I cannot be happier to announce a great and fun run from all of you. In addition, will reaching to all of you for input and help. this wonderful achievement from of my best the April 22 6th Annual TMII Symposium, which Meanwhile, please feel free to email me with colleagues and friends. You can read about his as usual has a superstar speakers line up: Bruce any thoughts on this topic or any topic. I wish journey with us at TMII and about some of his Fischl from MGH; Mark Griswold from Case you a great TMII newsletter read. accomplishments. Western; Anna Moore from MGH; Julie Price from the University of Pittsburgh; and Debiao Zahi Fayad, PhD As always we also have reasons to celebrate all Li from Cedars Sinai. Please don’t forget to Director, Translational & Molecular the progress being made by our wonderful TMII register for both events asap and also submit Imaging Institute members as exemplified by one of our recent your abstracts for the TMII Symposium. We look Professor of Radiology and Medicine trainees, Alan Seifert from the Gordon Xu’s forward for a continued robust participation. [email protected] WHAT’S NEW? TMII News & Updates Congratulations to Rebecca Feldman, PhD, view of the paper was published online in Willem Mulder, PhD. from Priti Balchandani’s High Field Imaging lab, Magnetic Resonance in Medicine. for her recent publication “A Semi-Adiabatic In addition to all the great conferences coming Lastly, the 2016 TMII calendar is now available Spectral-Spatial Spectroscopic Imaging (SASSI) up, don’t miss the ISMRM workshop on on Shutterfly. Thank you to all those in TMII who Sequence for Improved High Field Magnetic Molecular & Cellular MRI: Focus on Integration, contributed and Rebecca Feldman for putting it Resonance Spectroscopic Imaging”. An early June 8-11 in Amsterdam, chaired by TMIIs own, all together.

UPCOMING EVENTS

TMII Frontiers of Imaging Seminar Series > Thurs Feb 25, 2016 1 - 2pm: J. Thomas Vaughan, PhD - Professor, Univeristy of Minnesota “The future of MRI: More power for research and more utility for diagnostics” - Hess Center Seminar Room B

TMII Seminar Series > Tues Feb 9, 2016 12 - 1 pm: Satish Viswanath, PhD - Research Assistant Professor, Case Western Reserve University “TBA” - Hess TMII Conference Room s1-117 > Thurs Feb 11, 2016 11am - 12 pm: Joao Cavalcante, PhD - Assistant Professor, University of Pittsburgh “Risk Profiling in Aortic Stenosis Using Multimodality Imaging” - Hess Center TMII Conference Room s1-117 > Fri Feb 19, 2016 11am - 12pm: Naeim Bahrami, MS, PhD Canididate, Wake Forest University “DTI Imaging and structural network connectivity changes associated with Subconcussive Impacts in Youth Football Players“ - Hess Center TMII Conference Room s1-117 > Fri Feb 26, 2016 10 - 11am: Malgorzata Marjanska, PhD - Associate Professor, Univeristy of Minnesota “ltered Neurochemical Profile in the Healthy Elderly Brain Measured via 7 T 1H MRS” - Hess Center TMII Conference Room s1-117 > Thurs Apr 7, 2016 1 - 2pm: Elena Aikawa, MD, PhD - Associate Professor, Harvard Medical School/Brighman and Women’s Hospital “TBA” - Hess Center TMII Conference Room s1-117

> April 22, 2016 8am - 5pm: 6th Annual TMII Symposium - REGISTRATION AND ABSTRACT SUBMISSION OPEN! For more information on these and other events go to: http://tmii.mssm.edu/events

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | 1470 Madison Avenue | New York, NY 10029-6574 | tmii.mssm.edu FACULTY SPOTLIGHT Captivated by MR: TMII Faculty Awarded Full Professor and Tenure Willem Mulder, PhD

Captivated by an MR image of a circular After arriving in New York Dr. Mulder was Today, the Nanomedicine Lab resides at the structure on the cover of Circulation, one of the immediately productive, writing two reviews. seventh floor of Mount Sinai’s Hess Center top cardiovascular research journals, Dr. Willem Experimentally, he was very fortunate to get for Science and Medicine. The lab’s research Mulder decided to investigate the publication’s along very well with Dr. Fayad’s new postdoc efforts revolve around the development first author Zahi Fayad. At the time, early 2002, David Cormode, who now heads his own lab at and application of nanomedicine to Dr. Mulder freshly started his Ph.D. trajectory in UPenn. Together, diagnose and treat Professor Klaas Nicolay’s newly founded MRI lab they invigorated cardiovascular at the Department of Biomedical Engineering Dr. Fayad’s disease and cancer. of the Eindhoven University of Technology atherosclerosis Additionally, in The Netherlands. The initial focus of Dr. molecular the lab works to Mulder’s project was on MR phenotyping of imaging agent develop innovative atherosclerosis in mice. He soon discovered program, while technologies Dr. Fayad was clearly a leader in this field, after Dr. Mulder’s own to understand publishing a series of high profile papers on research focused nanoparticle in- vessel wall MRI in mice, rabbits and humans. on cardiovascular vivo behavior to Dr. Fayad was the rising star in cardiovascular nanomedicine. He translational studies imaging, which prompted Dr. Mulder to closely aspired to explore in pig models and monitor his lab’s output. the application nonhuman primates. of nanoparticle therapy to treat inflammatory The network is extensive and internationally Driven by atherosclerosis, for which his first student Mark oriented, formalized by Dr. Mulder’s increasingly Lobatto laid important groundwork. secondary Professor appointment at the growing Academic Medical Center (AMC) of the frustrations, In 2009, Dr. Fayad expressed his desire to apply University of Amsterdam. This arrangement the result of for the National Heart Lung and Blood Institute’s allows the exchange of students and doing research Program of Excellence in Nanotechnology to Dr. intellectual capital, exemplified by the first in a new lab Mulder at a dinner meeting in a SOHO restaurant. in human cardiovascular nanotherapy trials that still lacked They then outlined a strategy and put in an at the AMC; technology that was developed permits for work application for a $16.5 million program, which was in the Mount Sinai lab. in live mice, Dr. awarded to Dr. Fayad in 2010. This considerably Mulder decided elevated the nanomedicine effort and allowed “The nanomedicine team members make to exploit his undergraduate experience in Dr, Mulder establish a topnotch nano team at me particularly proud. They are diligent, therapeutic nanoparticle synthesis. Instead of Mount Sinai. At the same time, Dr. Mulder started collaborative and very hardworking trying to replicate Dr. Fayad’s work, he decided submitting his own R01 applications, the first one individuals, but they are also fun, to develop nanoparticle MRI contrast agents being awarded in 2011, the second and third in mischievous at times. Zahi [Fayad] taught that could be employed for vessel wall imaging. 2014. me to create a family-like mentality of trust, In retrospect, this was a decision that would generosity and loyalty. I took this advice determine Dr. Mulder’s scientific career. to heart and consider it one of the forces behind our success. I am very grateful for At a molecular imaging conference in Cologne the school’s support and its leadership’s (Germany) in 2005, Dr. Mulder had the pleasure vision, and feel the responsibility to keep on meeting Dr. Fayad in person. Dr Fayad invited elevating the level of innovation and quality him to spend a month in his lab later that year. of my lab’s output. To become a tenured “In April 2006, with a heavily sunburned head (I professor at one of America’s top medical didn’t realize a New York April sun could be this schools was the last thing on my mind when relentless) I met with Drs. Drayer and Charney, hugging my friends and family goodbye at and was offered the opportunity to establish Amsterdam Airport’s international terminal, my own lab by Dr. Fayad, who was recently a little over 9 years ago.” promoted to full professor. In October 2006

some of my family, friends and very recently Nanoparticles can be labeled with a variety of imaging acquired girlfriend Marielle, gathered at agents to enable their detection in plaque cells with CT, MRI, Willem J. M. Mulder, PhD Amsterdam Airport to escort me into this new optical methods, or nuclear imaging, such as PET and SPECT. Director, TMII Nanomedicine Lab The relative costs, sensitivity, scan time range, and resolution Professor of Radiology journey. “ range are indicated. [email protected]

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | 1470 Madison Avenue | New York, NY 10029-6574 | tmii.mssm.edu 2 SCIENCE SPOTLIGHT 6th Annual TMII Symposium

The TMII Symposium is a full day educational event where faculty, staff and trainees from Mount Sinai and outside institutions present The 6th Annual TMII Symposium their current or future research in the field of Friday April 22, 2016 Keynote Speaker Bruce Fischl, PhD medical imaging. The sessions throughout the Professor, Radiology Harvard Medical School/MGH Computational analysis of functional, connectional day are a mix of internationally recognized and architectonicproperties of the human brain with translational applications invited speakers and attendee-submitted Icahn School of Medicine Cancer/Body Imaging Nanomedicine Mark Griswold, PhD Anna Moore PhD poster and oral presentation. Each session, at Mount Sinai Professor, Radiology Professor, Radiology Case Western Reserve University Harvard Medical School/MGH Rethinking the way we do MRI: Image-guided Precision Nanomedicine cardiovascular imaging, neuroimaging, Magnetic Resonance Fingerprinting for Cancer Therapy cancer & body imaging and nano medicine, Hess Center for Science Cardiovascular Imaging Neuroimaging Debiao Li, PhD Julie Price, PhD and Medicine Professor, Biomedical Sciences Professor, Radiology will have one invited speaker give a talk Cedars Sinai Medical Center University of Pittsburgh MR Coronary Angiography and Vessel In vivo PET imaging of protein targets and one oral presentation chosen from the Wall Imaging in Alzheimer’s disease Registration Open submitted abstracts. Registeration and abstract tmii.mssm.edu tmii.mssm.edu/tmii2016 #TMII2016 submittions open now!

IMAGING SPOTLIGHT Cervical Spinal Cord and Brainstem Imaging at 7 T Alan C. Seifert, PhD - Xu Lab The spinal cord houses circuits which modulate inhomogeneities. TMII’s new 22-channel 7 T pain signals based on both descending input head and neck RF coil provides excellent B1+ from the brainstem and concurrent vibrotactile transmit efficiency and SNR throughout the stimuli from the periphery. Loss of modulatory entire brainstem and cervical spine, leading to input and preservation of sensory pathways improved image resolution. This new RF coil, may set the stage for neuropathic pain. Greater in combination with suppression of flowing understanding of these circuits would improve CSF, advanced shimming, and the enhanced our knowledge of chronic pain after spinal cord BOLD signal at ultra-high field will enable us to injury. investigate these subcortical structures with fMRI. Functional MRI is more difficult to perform in the brainstem and spinal cord than in the cortex because of the small size of these Alan C. Seifert, PhD structures and, due to their proximity to the Axial MEDIC images with 300 µm in-plane resolution from C1 to Xu Lab (Neuroimaging) heart, lungs, and vertebrae, their greater C7 vertebral levels, demonstrating excellent gray-white matter Postdoctoral Fellow vulnerability to motion and magnetic field contrast. Image courtesy of Joo-won Kim, PhD [email protected]

CORE SPOTLIGHT Siemens MAGNATOM Skyra 3T

This is an FDA approved 3 Tesla human MRI newly deigned RF system and coil architecture scanner. Its wide bore design (173 cm system integrates (Tim 4G) with all digital-in/digital- length with 70 cm) can accommodate subjects out technology. The scanner has an actively with larger body compositions compared to shielded water-cooled gradient system and the 60 cm bore of a typical clinical 1.5T & 3T. A zero helium boil-off. Specialized RF distribution increases uniformity in all body regions. On- board software is available for: neuro, angio, cardiac, body, onco applications. A variety of coils for all body parts and configuration are available including; a 32 channel head coil, 18 channel body matrix array, and an integrated 32 channel spine coil. fiber optic subject response gloves, pneumatic The TMII Skyra is equipped with the state of computerized headphones with microphones the art peripherals for functional imaging as well as a full spectrum of physiological including LCD goggles, integrated eye-tracking, recording probes for ECG, GSR, pulse-Ox etc.

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | 1470 Madison Avenue | New York, NY 10029-6574 | tmii.mssm.edu 3 BIC CORNER Please add two dates to your calendar for BIC User Workshops are recorded and can be website using the ‘Sign up’ button above the Brain Imaging Center (BIC) events. The first accessed from the website at https://bic.mssm. Twitter feed at https://bic.mssm.edu, or directly Annual BIC 10k run/walk/biking event will edu/events/bic-user-workshops/. Website from this link: https://bic.mssm.edu/accounts/ begin at 3:30pm on April 8 signup/. BIC’s resources and 2016, following the course development rely on user support. on the attached map. Over As NIH submission deadlines 40 of our colleagues have approach, please remember already registered- please the importance of including join by registering at https:// BIC in applications for funding. bic.mssm.edu/bic-1st-annual- A document with reference 10k-event/. Planning for the language for justifications of 3rd Annual BIC Symposium support is available for use in is underway. Please save grant preparations, at https://bic. the date- October 19 2016, mssm.edu/blog/including-bic-in- with Helen Mayberg as the upcoming-nih-grant-submissions/. keynote speaker. BIC has The website itself is constantly also begun the 2016 User undergoing development, so Workshop sessions, with presentations for content provides such technical areas only to please make use of it when searching for ‘How to Look at Your Data’ on January 12, and registered users who login. Members of the BIC information about brain imaging at Mount additional workshops being planned. The community can self-register directly from the Sinai.

CONTACTS Zahi A. Fayad, PhD Director, Translational and Molecular Imaging Institute Rafael O’Halloran, PhD Director, Cardiovascular Imaging Program Chief, Imaging Acquisition - BIC Faculty Professor of Radiology and Medicine (Cardiology) Assistant Professor of Radiology and Psychiatry [email protected] [email protected] Priti Balchandani, PhD Cheuk Y. Tang, PhD Director, High-Field MRI Program Director, Imaging Core Assistant Professor of Radiology and Neuroscience Associate Professor of Radiology and Psychiatry [email protected] [email protected] Prantik Kundu, PhD Bachir Taouli, MD Chief, Image Analysis Section & Advanced Functional Neuroimaging Section - BIC Faculty Director, Cancer and Body Imaging Program Professor of Radiology and Medicine Assistant Professor of Radiology and Psychiatry [email protected] [email protected] Venkatesh Mani, PhD Junqian Gordon Xu, PhD Director, Cardiovascular Imaging Clinical Trials Unit Neuroimaging Assistant Professor of Radiology and Neuroscience Assistant Professor of Radiology [email protected] [email protected] Willem J. M. Mulder, PhD Christopher J. Cannistraci, MS Director, Nanomedicine Program Program Manager Professor of Radiology Technical Operations Manager [email protected] [email protected]

Ways to keep in touch

Twitter: @TMIInyc Website: http://tmii.mssm.edu Facebook: TMII.SINAI Youtube: https://www.youtube.com/playlist?list=PLqLDR0CTP9_ Mailing Address: One Gustave L. Levy Place, Box 1234 otAZpwEy3EgOStthPo7V9f New York, NY 10029 Linkedin: https://www.linkedin.com/groups/Translational- Numbers: Tel: (212) 824-8466 Fax: (646) 537-9589 Molecular-Imaging-Institute-TMII-8358896/about

Icahn School of Medicine at Mount Sinai | Translational & Molecular Imaging Institute | 1470 Madison Avenue | New York, NY 10029-6574 | tmii.mssm.edu 4 TMII Human Imaging Core Dr. Cheuk Ying Tang, Dr. Lazar Fleysher, Dr. Johnny Ng, Edmund Wong, Daniel Samber, Victoria Wang, Chen Yang, Christopher Cannistraci PET/MR 7T Neuro-tesng room Skyra Spectral CT MR Simulator Clinical Exam room

TMII Neuro Funconal Imaging Peripherals Anesthesia Support TMII Server & Cluster in Data Center

Amyloid Imaging (AV45): PET-MR vs PET-CT

TMII Tracer (DTI-Tractography) Low-dose CT

Wholebody MRI Fusion: DTI – MRA - CT MRI SWI at 7 Tesla

Diffusion Spectrum Imaging TMII Connecvity Explorer TMII Pre-Clinical Imaging Core Dr. Cheuk Ying Tang, Yu Zhou, Dr. Lazar Fleysher and Dr. Johnny Ng, Victoria Wang, , Christopher Cannistraci

Bruker Micro MRI – 7T & 9.4T IVIS-Spectrum Opcal Imaging System Vevo2100 Micro-Ultrasound System

• Bruker Biospec 70/30USR with large bore B-GA 20S gradient • The IVIS Spectrum in vivo imaging system uses a novel patented opcal • B-Mode (2D) imaging for anatomical visualizaon and quanficaon, with imaging technology to facilitate non-invasive longitudinal monitoring of enhanced temporal resoluon with frame rates up to 740 fps (in 2D for a 4x4 (200mT/m, 640 T/m/s) and high performance gradient B-GA disease progression, cell trafficking and gene expression paerns in mm FOV) , and enhanced image uniformity with mulple focal zones. 12S (440mT/m, 3,440 T/m/s) living animals. • M-Mode for visualizaon and quanficaon of wall moon in cardiovascular research, single line acquision allows for the very high-temporal (1000 fps) • The IVIS Spectrum is a versale and advanced in vivo imaging system. An - Coils: 154mm CP resoluon necessary for analysis of LV funcon opmized set of high efficiency filters and spectral un-mixing algorithms • Pulsed-Wave Doppler Mode (PW) for quanficaon of blood flow - 4Ch Mouse Brain Phased Array lets you take full advantage of bioluminescent and fluorescent reporters • Color Doppler Mode for detecon of blood vessels including flow direconal - 4Ch Mouse Cardiac Phased Array across the blue to near infrared wavelength region. informaon and mean velocies; as well as for idenficaon of small vessels • - CP 35mm mouse body coil It also has the capability to use either trans-illuminaon (from the not visible in B-Mode boom) or epi-illuminaon (from the top) to illuminate in vivo • Power Doppler Mode for detecon and quanficaon of blood flow in small Bruker 7T - CP 1H/31P, 1H/13C, 1H/19F fluorescent sources. 3D diffuse fluorescence tomography can be vessels not visible in B-Mode; increased frame rates allow for significantly faster data acquision • performed to determine source localizaon and concentraon using the Bruker 9.4T 89mm vercal bore magnet with Micro combinaon of structured light and trans illuminaon fluorescent • Tissue Doppler Mode for quanficaon of myocardial ssue movement; for 2.5 Gradient System (2.5G/cm/A, up to 100G at 40A) images. example in assessing diastolic dysfuncon and MICWB40 In-Vivo micro imaging probe • The instrument is equipped with 10 narrow band excitaon filters • Vevo MicroMarker® Nonlinear Contrast Agent Imaging – for quanficaon of relave perfusion & molecular expression of endothelial cell surface markers; (30nm bandwidth) and 18 narrow band emission filters (20nm - 25mm & 30mm ID quadrature coils enhanced sensivity to Vevo MicroMarker contrast agents as linear ssue bandwidth) that assist in significantly reducing autofluorescence by the • signal is suppressed SA Isoflurane Anesthesia Setup spectral scanning of filters and the use of spectral unmixing algorithms. • 3D-Mode Imaging for anatomical and vascular visualizaon, when combined • SA Instruments Animal monitoring system In addion, the spectral unmixing tools allow the researcher to separate with either B-Mode, Power Doppler Mode or Nonlinear Contrast Imaging; Bruker 9.4T signals from mulple fluorescent reporters within the same animal. allows for quanficaon of volume and vascularity within a defined anatomical structure • ECG and Respiraon Gang are used to suppress imaging arfacts due to 10 excitation filters Spectral Unmixing 100 respiraon and cardiac movements. Both are important in cardiac and 80 abdominal imaging for both 2D and 3D data sets. 60

40

7T: Metabolic Syndromes (rsfMRI, DTI) Drs. Pasine, Wang, Tang 9.4T: Alzheimer’s (Hi-Res , 7T: Mulple Sclerosis, EAE Transmission% 20 Transducers:

beta Amyloid) Drs. Hof, Tang (T2W,DTI) Drs. Casaccia, Haines 0 *MS-200 12.5 or 21 MHz, Depth from 2mm to 36mm 400 440 480 520 560 600 640 680 720 760 Wavelength (nm) *MS-250 16 or 21 MHz, Depth from 2mm to 30mm 18 emission filters *MS-400 24 or 30 MHz, Depth from 2mm to 20mm IVIS-Spectrum System *MS-550D 32 or 40 MHz, Depth from 1mm to 15mm

VEVO 2100 Micro-Ultrasound System 7T:Neovascularizaon - stroke (MRA, DWI, DCE) Drs. Georgakopoulos, Tang

7T: 19F Spectroscopy of 19F, Drs. Fleysher, Marnez, Goldstein, Tang,

Image from Perkinelmer

Opmal Transilluminaon for Breast Cancer Firefly Luciferase Images Deep Tissue Sources Automac Mouse Atlas B-Mode and Color Doppler Mode M-Mode Pulsed-Wave Doppler Mode (PW) Transillumination Registraon in LI4.0 from Dr. Mihaela Skobe’s Lab Epi Illumination

7T: Lung Tumor (T2W) Dr. Goutham Narla 7T: Prostate Cancer (T2W,DWI) Drs. Pisipa, Tewari , Tang, Taouli

Power Doppler Mode Tissue Doppler Mode Image-Guided Needle Injecon

Bioluminescent image Aorta of Rabbit Cy5.5 sig/ bkg=1.43 sig/ bkg=90.51 with Firefly luciferase Image from Mark Unregistered Coregistered Image from Perkinelmer 9.4T: Teeth as archives of exposure (T1W-T2W) Drs. Arora, Mani 7T: permeability in ApoE-/- atheroscleroc mice , Drs. Fayad, Calcagno tumor cells in 96 wells Lobao (TMII) Image from Perkinelmer plate from Zewei Jiang (Dr. Samir Parekh’s lab) 3D-Mode Imaging Vevo MicroMarker® Nonlinear VevoStrain™ Analysis soware Contrast Agent Imaging for cardiac research Ancillaries

7T: Canine brains (hi-resT1, DSI) Drs. Spocter, Tang, Hof Gamma Counter Centrifuge Shielded Fume Hood Dose Calibrator Procedure Room Procedure Room (radioacve)