Principles of MRI EE225E / BIO265

Instructor: Miki Lustig UC Berkeley, EECS Today.... • Administration – http://inst.eecs.berkeley.edu/~ee225e/sp16/ • Intro to Medical Imaging and MRI

M. Lustig, EECS UC Berkeley Medical Imaging (Before 1895) • Only way to see is to cut!

M. Lustig, EECS UC Berkeley Medical Imaging (Post 1895) • Revolutionized diagnostic medicine • See internal anatomy • Visualize function • Many modalities • Many sources of contrast

M. Lustig, EECS UC Berkeley Basic Concept

Energy Energy Body Source Detection

Imaging System (Electronics, control, computing, algorithms, visualization....) e.g., Engineering!

M. Lustig, EECS UC Berkeley Medical Imaging System Requirements

• Diagnostic contrast • Inexpensive • Sensitivity • Easy to use • Specificity • Function • Can’t satisfy all • High Spatial-resolution • Many modalities • High Temporal-resolution • Often several used to • Safe make diagnosis • Fast

M. Lustig, EECS UC Berkeley Common Imaging Modalities

• Projection X-Ray (Electromagnetic)

• Computed Tomography (Electromagnetic)

• UltraSound (Sound waves)

• Positron Emission Tomography (Nuclear)

• Single-Photon Emission Tomography (Nuclear)

• Magnetic Resonance Imaging (magnetic)

M. Lustig, EECS UC Berkeley Medical Imaging is Multi-Disciplinary

Math Medicine

Engineering

Physics Biology

Chemistry

M. Lustig, EECS UC Berkeley Engineering Advances 1st x-ray (1895) x-ray today

M. Lustig, EECS UC Berkeley History: X-Rays History: Computed Tomography

Wilhelm Conrad Röntgen The breakthrough: 2 8 November 1895: discovers X-rays. 2 acquiring many projections around the object enables the reconstruction of the 3D 2 22 November 1895: X-rays Mrs. RöntgenCs hand. object (or a cross-sectional 2D slice) 2 1901: receives first Nobel Prize in physics CT reconstruction pioneers: An early X-ray imaging system: 2 1917: Johann Radon establishes the mathematical framework for tomography, now called the Radon transform. 2 1963: Allan Cormack publishes mathematical analysis of tomographic image reconstruction, unaware of RadonCs work. 2 1972: Godfrey Hounsfield develops first CT system, unaware of either Radon or CormackCs work, develops his own reconstruction method. 2 1979 Hounsfield and Cormack receive the Nobel Prize in Physiology or Medicine. Note: so far all we can see is a projection across the patient: Radon Cormack Hounsfield

Computed Tomography: Concept Computed Tomography: Past and Present

Image from the Siemens Siretom CT scanner, ca. 1975

2 128x128 matrix. Engineering Advances

early CT (1975) CT today

M. Lustig, EECS UC Berkeley

Modern CT image acquired with a Siemens scanner 2 512x512 matrix Cartotid Stenosis Virtual Medicine

Virtual colonoscopy, endoscopy, arthroscopy Virtual therapy and surgery planning Training platform

History: Ultrasound Ultrasound: Present

1942: Dr. Karl Theodore Dussik, @ transmission ultrasound investigation of the brain 1955: Holmes and Howry @ Subject submerged in water tank to achieve good acoustic coupling image of normal neck 3D Ultrasound

Engineering Advances early ultrasound (1959) ultrasound today

1959: Automatic scanner, Glasgow

MRI today

Intravasular ultrasound M. Lustig, EECS UC Berkeley Doppler ultrasound twin gestation sacs (s) and bladder (B). 1978, British Journal of Radiology, 51, 921-922 NOVEMBER 1978

Short communications Human whole body line-scan imaging by NMR By P. Mansfield, I. L. Pykett, and P. G. Morris Department of Physics, University of Nottingham and R. E. Coupland Department of Human Morphology, School of Medicine, University of Nottingham {Received July, 1978)

Historically, the first nuclear magnetic resonance (NMR) fast line-scanning adaptation (Mansfield et al., 1976) is the pictures demonstrating live human anatomy were produced basis of the technique reported in this paper. The thickness recently (Mansfield and Maudsley, 1976; 1977) using a line- of the cross-sectional slice examined was set to approxi- scanning technique. The original specimens studied were mately four cm in these first experiments. Selected lines of limited in object size to cross-sectional views through material across the slice are scanned from one edge of the fingers. A number of other approaches to imaging by NMR subject to the other and the subsequent data sets for each are also being developed and are compared, discussed and scanned line are stored in a computer for subsequent display referred to elsewhere (Brunner and Ernst, 1978; Pykett and as a picture (Baines and Mansfield, 1976). Large signals Mansfield, 1978). which produce the bright zones in the images come in Recent development and expansion of our imaging general from high concentrations of mobile protons con- apparatus has now allowed us to examine objects of the tained within or near the various tissues and organs. By dimensions of the whole human body. In this paper we wish mobile protons, we have in mind those associated with the to report the first NMR line-scan medical image of a thin free water, fat or oil contained in or distributed throughout cross-sectional slice of a live human body. the various tissues and organs of the body. Our work should be seen in the general context of other The whole body image shown in Fig. 1 is a transverse medical imaging schemes (Hounsfield, 1973; Hill, 1976; section through the abdominal region (of PM) at the level of Budinger, 1974), where improvements in techniques and the third and lower part of the second lumbar vertebrae. picture quality have recently achieved very high standards. Figure 2 is a line tracing of Fig. 1 labelling the recognizable NMR imaging, however, is in its infancy, but nevertheless morphological features. The liver is visible as a well-defined has merit as a non-invasive and, we believe, non-hazardous mass mainly on the right of the picture and blends with the technique. There is also the possibility of using NMR imag- abdominal wall at the periphery. The vertebrae are identi- ing in conjunction with other measurable NMR parameters fiable in the mid-line and the kidneys lie in the mid-lateral for Engineeringnormal and abnormal Advancestissue characterization and for region with the spleen adjacent on the left. The central physiological studies including blood flow measurements. circular images probably correspond to the abdominal aorta, Our NMR imaging technique, based on selective excita- with the pancreas lying anterior, inferior vena cava and duo- tion in switched magnetic field gradients, was first described denum. We believe that the variable dense anterior central by Garroway et al. (1974). A slightly modified version of the early MRI (1978) MRI today

FIG. 1. FIG. 2. Cross-sectional line-scan NMR image through the abdomen Labelled image of Fig. 1. A = aorta, C = colon, D = duo- at L2-3. Arrow indicates mid-line posterior. Left side lies to denum, G = gall-bladder, I = inferior vena cava, K = the left of the illustration. Bright zones correspond in general kidneys, L = liver, P = pancreas, S = spleen, SI = stomach to high mobile proton content. See Fig. 2 for labelled and intestines, V=vertebra. Abdominal muscles and retro- details. peritoneal fat (MF) are seen adjacent to the vertebra.

M. Lustig, EECS UC Berkeley 921 Projection X-Ray • Projection Format Pneumonia • Small Dose • Fast • Inexpensive

Projection

M. Lustig, EECS UC Berkeley Computed Tomography (CT) abdominal CT • Tomographic • Fast • High-Res • Moderate dose • ~1M$

Many Projections

M. Lustig, EECS UC Berkeley Computed Tomography

Sinogram cross-section

FBP

x-ray source

M. Lustig, EECS UC Berkeley Computed Tomography • Gantry rotation

M. Lustig, EECS UC Berkeley http://www.youtube.com/watch?v=4gklQHM19aY&feature=related Ultrasound Echo • Real-time • Inexpensive • No-radiation • Many applications

• Low contrast and penetration

M. Lustig, EECS UC Berkeley Anatomy vs Function

M. Lustig, EECS UC Berkeley Nuclear Medicine brain metabolism • Specific metabolic information (function) • Low-res • High dose • 1-2M$ PET

• SPECT: Gamma radiation • PET: Positron-> Gamma

M. Lustig, EECS UC Berkeley Magnetic Resonance Imaging (MRI)

• NMR: Nuclear Magnetic Resonance • MRI : Magnetic Resonance Imaging – please don’t say MRI imaging!

• MRI is VERY VERY VERY different from CT! • Cost: 1M-3M, mainly because of the Magnet

M. Lustig, EECS UC Berkeley History

• 1946 - Felix Bloch (Stanford) Edward Purcell (Harvard) independently discovered NMR. Nobel Prize (Physics) in 1952. • 1971 - Raymond Damadian showed changes in MR parameters (T1 and T2) in cancer. People started thinking about medical NMR applications. • 1972 - Invention of CT by Hounsfield and Cormack. Nobel Prize (Medicine) in 1979. • 1973 - Lauterbur described MRI in a similar way to CT

M. Lustig, EECS UC Berkeley History • 1975 - Ernst proposed key concepts. Nobel prize (Chemistry) 1991. • 1970’s - Mansfield contributes key ideas (slice selection) • 1982 - Widespread clinical MRI begins. • 2003 - Lauterbur/Mansfield receive Nobel prize (Medicine) for their contributions.

M. Lustig, EECS UC Berkeley MR Imaging • Magnetic resonance imaging has revolutionized medicine • Directly visualizes soft tissues in 3D • Wide range of contrast mechanisms – Tissue character (solid, soft, liquid, fat, ...) – Diffusion – Temperature – Flow, velocity – Oxygen Saturation

M. Lustig, EECS UC Berkeley Neuro Examples

PD T1 T2

Many different contrasts available

K. Pauly, G. Gold M. Lustig, EECS UC Berkeley Stanford Rad 220 Clinical Example PrecontrastPrecontrastNo Contrast Agent PostcontrastPostcontrastContrast Agent

M. Lustig, EECS UC Berkeley*K. Pauly, G. Gold, RAD220 Body Examples

Abdominal Blood Vessels Knee

M. Lustig, EECS UC Berkeley*K. Pauly, G. Gold, RAD220 Cardiac MRI

gated real-time

M. Lustig,*Courtesy EECS UC Juan Berkeley Santos M. Lustig, EECS UC Berkeley Flow Imaging Examples 4D flow

Real-time color flow

*Juan Santos, Stanford M. Lustig, EECS UC Berkeley *Marc Alley, Stanford Diffusion Examples

T2 weighted standard MRI White matter fibers 3 hours after a stroke

diffusion weighted MRI diffusion tensor imaging 3 hours after a stroke

*Dr. Steven Warach, Beth Israel Hospital, Boston, MA M. Lustig, EECS UC Berkeley *The Virtual Hospital (www.vh.org); TH Williams, N Gluhbegovic, JY Jew *Brian Wandell, Stanford Functional MRI Example

Sensitivity to blood oxygenation - response to brain activity

*Karla Miller, Oxford *BrianM. Lustig, Wandell, StanfordEECS UC Berkeley Taking fMRI further • fMRI decoding : “Mind Reading” Gallant Lab, UC Berkeley

M. Lustig, EECS UC Berkeley Spectroscopy Imaging • Functional Imaging (metabolism) • Also other nuclei (13C, phosphor)

Tumor

pyruvate lactate

* P. Larson, D Vigneron, UCSF M. Lustig, EECS UC Berkeley Go Bears!

M. Lustig, EECS UC Berkeley The data is complex: there is magnitude and phase. We typically look at the magnitude.

Phase Magnitude Phase

*K. Pauly, G. Gold, RAD220 Fat on Gradient Echo Images

water and fat are “in phase” “ out of phase” Breast

*K. Pauly, G. Gold, RAD220 Spin-Echo vs Gradient Echo Gradient Recalled Echo Spin Echo

48 *K. Pauly, G. Gold, RAD220 Chemical-shift

*images, courtesy of Brian Hargreaves Motion Artifacts

Pelvic Scan after patient became Abdomen Scan (Good Quality) restless (non-diagnostic) ` • If you have an iPhone, please install: – Tone Generator app (Michael Heinz)

• How Does MRI Work? • Magnetic Polarization -- Very strong uniform magnet • Excitation -- Very powerful RF transmitter • Acquisition -- Location is encoded by gradient magnetic fields -- Very powerful audio amps

M. Lustig, EECS UC Berkeley Polarization • Protons have a magnetic moment • Protons have spins • Like rotating magnets

M. Lustig, EECS UC Berkeley Polarization

• Body has a lot of protons • In a strong magnetic field B0, spins align with B0 giving a net magnetization

M. Lustig, EECS UC Berkeley *Graphic rendering Bill Overall Polarizing Magnet

• 0.1 to 12 Tesla • 0.5 to 3 T common • 1 T is 10,000 Gauss • Earth’s field is 0.5G • Typically a superconducting magnet B0

M. Lustig, EECS UC Berkeley Typical 1.5T MRI System

M. Lustig, EECS UC Berkeley Polarizaion • Polarization results in net magnetization

M. Lustig, EECS UC Berkeley Free Precession

• Much like a spinning top • Frequency proportional to the field • f = 64MhZ @ 1.5T

MIT physics demos

M. Lustig, EECS UC Berkeley Free Precession

• Precession induces magnetic flux • Flux induces voltage in a coil

Signal

M. Lustig, EECS UC Berkeley Intro to MRI - The NMR signal

• Signal from 1H (mostly water) B0 • Magnetic field ⇒ Magnetization • Radio frequency ⇒ Excitation • Frequency ∝ Magnetic field

time frequency

γB0

M. Lustig, EECS UC Berkeley Intro to MRI - The NMR signal

• Signal from 1H (mostly water) B0 • Magnetic field ⇒ Magnetization • Radio frequency ⇒ Excitation • Frequency ∝ Magnetic field

time frequency

γB0

M. Lustig, EECS UC Berkeley Intro to MRI - Imaging

• B0 Missing spatial information B0

center

time frequency

γB0

M. Lustig, EECS UC Berkeley Intro to MRI - Imaging

• B0 Missing spatial information G B0 • Add gradient field, G

Weaker center field Stronger unchanged field time frequency

γB0

M. Lustig, EECS UC Berkeley Intro to MRI - Imaging

• B0 Missing spatial information G B0 • Add gradient field, G • Mapping: spatial position ⇒ frequency

center

time frequency

Fourier

Transform

Reference (γB0)

M. Lustig, EECS UC Berkeley iPhone Imaging

M. Lustig, EECS UC Berkeley Imaging geometry

3D Volume 3D Slab 2D Sections Simple Slice Selective Imaging

Excitation RF

A/D Slice selection

z ω = γ Gzz

Slice Profile Slice Frequency ω

RF Spectrum Simple Slice Selective Imaging

Gz Acquisition Gy

A/D 2D Imaging: k-space

• Gradient fields cause k-space image space linear phase

• Receive integrated signal

• give spatial frequency information

max “velocity” max “acceleration” 2DFT, or spin warp

Gx A/D

• One line of k-space for kx each acquisition • Repeat N times for an NxN image • Recon is 2DFT • TR 400 ms, 256x256 image < 2 minutes Pulse Sequence 2DFT pulse-sequence Fourier MR Imaging

k-space (Raw Data) Image

Fourier transform

M. Lustig, EECS UC Berkeley Pulse Sequence Spiral pulse-sequence k-space Sampling - resolution k-space Sampling - FOV

Δk=1/FOV Δk=1/FOV

FOV MRI is all about contrast......

http://thundafunda.com Relaxation xy

e-t/T2 Time Transverse M Transverse z 1-e-t/T1

Time Longitudinal M Longitudinal

• T2 = Active flushing ~5 second

• T1 = Refilling time ~1min

M. Lustig, EECS UC Berkeley The Toilette Analogy, Steady-state • Flush - Refill

• Flush continuously – Never fully refills – After a while, same from flush to flush – “Steady state” • Timing creates contrast

M. Lustig, EECS UC Berkeley Contrast T1 T2

M. Lustig, EECS UC Berkeley House Prefers T2 .....

You -- Get cervical, thoracic and lumbar T2 weighted Fast Spin-Echo MRIs

M. Lustig, EECS UC Berkeley MRI at UCSF and Berkeley

• Berkeley has a 3.0 T Siemens whole body scanner dedicated to Neuroscience at the Helen Wills Neuroscience Institute • UCSF has several GE & Siemens scanners (1.5T, 3T and a 7T) • Berkeley has Prof. Alex Pines, John Clarke, Tom Budinger, Steve Conolly and myself on the basic NMR/MRI technology • UCSF has dozens of MRI researchers Peder Larson, Daniel Vigneron, Xiaoliang Zhang, Duan Xu, John Kurhanewicz, Sarah Nelson, Sharmila Majumdar, David Saloner, Nola Hylton, Michael Wendland, Sabrina Ronen, and Alastair Martin