Quantitative Monitoring of Mouse Lung Tumors by Magnetic Resonance

Home , Nature Protocols

© 2012 Nature America, Inc. All rights reserved. imaging (BLI) of tumor cells expressing a marker gene such as as such gene marker a expressing cells tumor of (BLI) imaging scanning (CT) tomography computerized as such modalities imaging nondestructive therapy.to of use The intracavitary malignancies, such as lung cancer, and their response slides histological or blocks lung on nodules as killing animals and analyzing tumor burden by counting tumor for and treating strategies lung therapeutic cancer biological studying for system reliable and reproducible a offered the furthered range of experimental tools substantially available for has modeling lung cancer p53) and of loss-of-function K-ras, neered mouse models of lung cancer (e.g., activation of oncogenic growth tumor lung of cancerinjection cell lines, which result in intrapulmonary administration of carcinogen the either on relied have lung) the within occurring tumors (i.e., sites subcutaneous and lung the between immunosurveillance in differences derable logy of tumors that develop and grow in the lung because of consi physio the to directly translate not does it tissue, soft or skin the meter dia tumor of measurement direct the by evaluated is therapy to rely on the subcutaneous injection of tumor cell lines, and response of such a therapy related preclinical small-animal studies demonstrating the success melanoma malignant of treatment for clinical success in the blockade of cytotoxic T lymphocyte antigen-4 trials. clinical For beforeoptions peutic initiating example, recent thera testing and malignancies of understanding our advancing and morbidity society of our in source mortality major a remain tumors Pulmonary I evaluating new compounds targeting lung cancer tool for preclinical investigation of therapeutic strategies for treatment of pulmonary malignancies, and it may be valuable in primary (discrete tumors) and metastatic (diffuse tumors) disease. 1 h (data acquisition plus analysis) per mouse. Quantitative, analytical methods are described for measuring tumor burden in both segmentation algorithms described by our laboratory, pulmonary tumor burden can be quantitatively measured in approximately and serial monitoring of pulmonary tumors. By combining respiratory-gated data acquisition methods with manual and automated in situ of improved therapeutic strategies. However, detection of mouse pulmonary tumors and their subsequent response to therapy recurrence of extrathoracic malignancies. P Published online 5 January 2012; doi:10.1038/nprot.2011.424 Milwaukee, Wisconsin, USA. St. Louis, St. Louis, Missouri, USA. 1 Haris G Vikis Alexander Sasha Krupnick magnetic resonance imaging Quantitative monitoring of mouse lung tumors by 128 ography (PET) scanning, has also been reported luciferase Department of Surgery,of Department Washington University in St. Louis, St. Louis, Missouri, USA. NTRO protocol rimary rimary lung cancer remains the leading cause of cancer-related death in the Western world, and the lung is a common site for Many investigators have relied on destructive modalities, such such modalities, destructive on relied have investigators Many Traditional models for the study of orthotopic mouse lung cancer

| VOL.7 NO.1VOL.7 6, is challenging. We have recently described M D 7 . this Although approach may be valid for of malignancies UCT 2 2 , such or imaging as metabolic positron emission tom I ON 11,1 4 | 2012 , Andrew E Gelman 6 2 . Most of these preclinical experiments, however, . The recent development of genetically engi genetically of development recent The . 8, 9 . |

5 natureprotocols These These authors contributed equally to this work. Correspondence should be addressed to J.R.G. ([email protected]). 1– 3 . Mouse models have a crucial role in in role crucial a have models Mouse . 3 Department of Radiology,of Department Washington University in St. Louis, St. Louis, Missouri, USA. 1 0 or intratracheal or intravenous (i.v.) 1 , 5 , Vanessa K Tidwell 4, 1 5 , Daniel Kreisel can be linked directly to directly linked be can S mall-animal mall-animal (rodent) models of cancer can have a very valuable role in the development 11,2 14,2 0 , bioluminescent bioluminescent , 2 1 3 , for monitoring monitoring for , ( 13–2 in in vivo Table 0 R . I I as a reliable, reproducible and nondestructive modality for the detection 2 , 5 1 . , John A Engelbach 1 ). & Joel R Garbow - - -

2 Department of Electrical and Systems Electrical of Engineering, Department Washington University in role in preclinical small-animal studies of tumor therapy tumor of studies small-animal preclinical in role tions, all three of these haveimaging modalities had an (ref. important mm 1–2 is PET small-animal ( low is both modalities of resolution anatomic the express to (bioluminescence), engineered luciferase tumors of detection by or (PET) tracer mation regarding tumor burden, either by uptake of a radioactive infor provide can BLI and PET scanning both Although studies. studies of lung cancer. lung of studies small-animal for advantageous be may line, cell tumor a of tion manipula extraneous nor tumor the into genes foreign of tion resolution that avoids ionizing radiation, and that requires anatomic neither introduc high with modality imaging an However, response pr As homeostatic T cell system immune the repopulate to expand they as phocytes lym T native lead remaining the of (5–6 proliferation lymphopenia, Gy homeostatic to ing peripheral 2.7–4.2 in as result can low scans) as micro-CT irradiation whole-body of Doses studies. drug long-term to designing when be taken into account CT on scans need serial system the of immune effects cumulative the addition, In considered. be must imaging, serial with cially espe growth, tumor affect can micro-CT by delivered radiation cancer lung treat to used typically doses therapeutic the than lower log-fold several are doses radiation from in 0.22 to doses ranging 0.76 Gy per scan toring of the disease and have used this for technology both detection and serial moni cancer lung of models mouse to micro-CT adapted have groups nodules tion, speed of data and to acquisition ability detect subcent resolu high its of because in humans cancer lung of monitoring CT is of choice the modality for imaging and screening, diagnosis Computerized tomography T hus, small-animal M 24,2 3 3 3 , even low doses of mayradiation confound therapeutic 3 5 , Vamsi V Alli . On the basis of this success in humans, multiple multiple humans, in success this of basis the On . Table 11,2 R 6 I I represents a novel and unique research . However, micro-CT can deliver radiation o 1 liferation can lead to a tumor-immune to a tumor-immune lead can liferation ) 1 23,3 , Arye Nehorai 4 Medical College of Wisconsin Cancer Center, 4 (e.g., typical spatial resolution of of resolution spatial typical (e.g., 35)). Despite these limita these Despite 35)). 28,2 9 , the possibility that that possibility the , 2 , Ming You 20,2 7 . such Although 4 i , meter meter

20,23,2

30–3

6 2 ------. .

© 2012 Nature America, Inc. All rights reserved. molecules in different tissues and organs often have different relaxato sufficient article), is knowthis waterit the that of purposes the textbooks MRI and NMR standard of variety a in found be can relaxation, T2 and T1 ing of signal following RF excitation. Further details about MRI, includ(RF) excitation, and transverse (T2) relaxation, describing the loss the return to equilibrium of the magnetization after radio frequency acteristic relaxation times: longitudinal (T1) relaxation, describing resonance,In magnetic spins are often characterized by two char present in the sample and on the relaxation properties of that water. in water. The observed signal depends on both the amount of water fromhydrogen signal nuclei involveof studies detection MRI the preclinical and clinical of majority vast The methods. analytical destructive other with possible not are that studies (time-course) nondestructive technique that enables a wide variety disease. of longitudinalof models animal and MRI is a powerful tool imaging for animal systemscharacterizing resonance Magnetic imaging T tion times, as do healthy and pathologic tissues. These a Micro-CT MRI Modality Optical Micro-PET b le 1 |

Comparison Comparison of modalities for noninvasive expression with fluorescence imaging (FI) and Image-specific metabolic pathways and protein burden using BLI High-throughput determination of tumor or portable instrumentation Relatively inexpensive, compact benchtop Subject not exposed to ionizing radiation Quantitative measurement of uptake kinetics High signal detection sensitivity systemic metastasis Metabolic evaluation of primary tumor and tumor and healthy lung tissue Excellent endogenous image contrast between High throughput (~10 min per subject) technology, but preferred for high resolution No need for respiratory gating with free-breathing High spatial resolution contrast-enhanced imaging Visualization of vasculature using dynamic changes in spatial and temporal tissue morphology High spatial resolution, with ability to quantify tumor and healthy lung tissue Excellent endogenous image contrast between Subject not exposed to ionizing radiation S fluorescent reporters within the same animal High signal detection sensitivity of multiple fluorescent molecular probes trengths 36–3 8 . For the practical use of MRI (and (and MRI of use practical the For . In vivo In MRI is a noninvasive and and noninvasive a is MRI in in vivo

differences in

imaging of pulmonary tumors.

- - - Requires synthesis and administration of Low spatial resolution, lack of anatomic information Subject exposed to ionizing radiation contrast agent Visualization of vasculature requires a large dose of Requires access to a small-animal CT scanner Subject exposed to ionizing radiation Not high throughput (~40 min per subject) clinical scanner or small-animalpurpose-built RF coil for Requires access to a small-animal MRI scanner Requires respiratory gating Weaknesses metabolic pathway or enzyme activity fluorescence-tagged reporter probes to visualize protein or administration of tumor-selective FI requires expression of gene-encoded fluorescent (luciferase) by tumor cells BLI requires expression of biomarker lack of anatomic information Low spatial resolution at depths Requires access to a small-animal PET scanner radiolabeled tracers both magnetic susceptibility to and due static field signal inhomogeneity of loss the characterizes T2* inhomogeneity; field susce magnetic to due signal of decay the characterizes (T2 ratios lowsignal-to-noise to contributing further thus constants, time relaxation T2 and T2* air- tissue many interfaces the of the alveoliwith and associated bronchioles, whichsusceptibility result magnetic in short in ations vari (ii) ratio; signal-to-noise the limits severely which lung, the to lung MRI are (i) low tissue density and low water content within methods innovative and new of ment develop the required which MRI, for challenges unique present however,Lungs, tissues. and organs other many in malignancies brain primary of study the for developed well been has tissue, soft in necessary to generate high-resolution MR images. either T1 or T2 (or both) can be exploited to provide the contrast blurring in the absence of motion-synchronized data motion-synchronized in absence the of blurring (iii) respiratory and cardiac motions, which lead to significant image MRI signal intensity, which depends on the density of water water of density the on depends which intensity, signal MRI 3 9 , prostate 4 0 and liver tumors p tibility in the absence of static magnetic magnetic static of absence the in tibility natureprotocols

500

4 µ 1 , as well as for the evaluation of m,

42–4

4 . The unique challenges challenges unique .The

| VOL.7 NO.1VOL.7 protocol R 23,48,54,55 23,27,48,53 46,48,51,52 23,42,43,45, 56–60 14,23,46, eferences

acquisition. | 2012 3 6

); and |

129

- -

© 2012 Nature America, Inc. All rights reserved. well-defined, discrete tumors within the lungs. Typically, pri Typically, lungs. the within tumors discrete well-defined, in to lung cancer one mice or leads of the presence more Primary (discrete) lung tumors primary of Manualcalculation segmentation/volume are provided. ( ( equipment large selected preferences.of andPhotos needs gator’s individual investian accordingto modified However,be can equipment all preferences.laboratory’s our represents MATERIALSsection the Comment regarding the equipment. pulmonary malignancies. cancer,thereby treatmentsadvancingnew developmentfor the of able for the characterization of mouse experimental models of lung invalu be proveto will manuscript this in protocols the total,In priate shape and signal models for healthy organ tissue and tumor. although this would require the (nontrivial) development of appro herein couldlikely extended be described to other organsystems (e.g., liver, brain), method semiautomated The lung. healthy of tumors appear as bright features against a dark image background development automatedof methods to estimate tumor burden, as lung tumor burden lung MR images that allows for the rapid estimationmetastatic of validated a nearly fully automated algorithm for segmenting mouse lungs both or one in chyma which tumor tissue replaces a large portion healthyof lung paren limit is a diameter of ~0.5 mm) and extensive tumor metastases, in tion of both individual, submillimeter lesions (minimum detection our group in multiple settings for identification and serial examina lung, using small-animal MRI. These protocols have been used by burden, bothlung of primary cancer and metastatic disease to the tumors and path lung of characterization the for opportunities the Nevertheless, bearing mice, following the instructions provided in Step 54C. and automatic lung segmentations of four control mice and ten tumor- tumor burden. The calibration curve was generated based on lung weights Figure 1 130 diameter. in mm 4.0 to 0.5 from size in range tumors lung mary Supplementary Fig. 2 protocol

Here we present a protocol for quantifying pulmonary tumor pulmonary quantifying for protocol a present we Here MRI-measured tumor burden (arbitrary units)

| 1 2 3 4 5 6 VOL.7 NO.1VOL.7 × × × × × × 10 10 10 10 10 10

| 0 4 4 4 4 4 4 Sample calibration curve for the estimation of absolute lung 0 Supplementary Fig.1 Supplementary | 2012 o 100 logy by logy MRI are substantial. 4 9 . The lung is a particularly favorable case for the | ) required for the acquisition of MR images natureprotocols Corrected tumorweight(mg) 200 42,43,45–4 8 ) and small-animal equipment small-animal and ) 300 . We have further defined and defined .We further have

The equipment described in 400 500 600 ------

lung lung weight. Todif account experiment-to-experiment for small lung image intensity and tumor burden, as derived from measured Figure burden tumor lung total of measure accurate an provides sity inten image lung MR average tumors, diffuse with mice for that for assessing total tumor burden is needed. We method havealternate ROIs,an recentlyindividual by shown circumscribed being to and disease, are not in amenable tumors which diffuse metastatic multiple well-defined, discrete tumors are observed. However, for measuring tumor volumes in mice with tumors,primary in which The techniques described above are appropriate for quantitatively lung (diffuse) tumors metastatic of Manualcalculation segmentation/volume animal. that within tumor individual each of volume the of sum the simply then is mouse given a for burden tumor total contiguous slices corresponding to the same common tumor. The from volumes together adding and by thickness slice the tiplying areas the calculating an cross-sectional within slice,imaging mul drawing nodules, observed by the of each around (ROIs) calculated interest of regions be can tumor each lung for mice, volumes these In tumor tissue. lung healthy of image background dark the against spots bright as identified be can tumors These requires the development of a calibration curve, as detailed in Tidwell that of a cohort of control mice that are not tumor-bearing. This measurement (in units of mg) by comparing average, normalized lung image intensity with (left) and light (right) tumor burden. ( relative lung tumor burden between mice; shown here are animals with heavy response to therapy in individual mice via serial imaging. ( tumor burden. ( Figure 2 in illustrated mice, of as a burden across series tumor lung absolute measuring mice; and (iii) of relative burden a lung across tumor series uring monitoring tumor progression or therapeutic response); (ii) meas tumor burden in the same mouse lung at two qualitatively, or more time points assessing, (e.g., (i) cases: different three separately discuss and distinguish to useful is it measurements, these ering consid In animal. each within liver the of that to normalized is tuning or animal positioning, the average coil MR RF lung in image intensity variations to due intensity image absolute in ferences b c a Absolute tumorburdencomparedwithcontrol(nottumor-bearing) Relative lungtumorburdenindifferentmice Progression ofdiseaseinthesamemouse lungs Normal

1 | Liver Graphic illustrating three different approaches to measuring diffuse is a plot showing the correlation between average MR MR average between correlation the showing plot a is a Figure Figure ) Monitor the progression of disease (as illustrated here) or 2 . Tumors c ) Measure the absolute tumor burden b ) Compare the et al. 4 4 9 . 9 - - - - - . © 2012 Nature America, Inc. All rights reserved. • • • hardwareMRI analysis andsoftware • • • • • • • • • Miscellaneous supplies • • • Anesthesia • • • • • • • • MRI acquisition equipment EQUIPMENT • • • • • • REAGENTS M algorithm new a validated also tumors lung measuring quanti tatively for intensity image lung average of utility the showing automatically and the segment the lungs throughout mouse. Our recent accurately publication to ability the by aided greatly be would procedure analysis the lungs. of the efficiency and of throughput The segmentation is step labor- the time-consuming intensity, intensive, image lung mouse average calculating In lung (diffuse) tumors Automated metastatic of calculation segmentation/volume The programs have programs The successfully also been tested versions inearlier of using MATLAB version 7.6.0, R2008a, theImage Processing with Toolbox. MATLAB (MathWorks) Code wasdeveloped andcalculationsperformed National Health Institutes (NIH); of available athttp://rsbweb.nih.gov/ij/) ImageJ software (Java-based, public-domain software developed attheUS no. PC(model Dell 390) strainer Cell Tygon tubing, ½inch North (Saint-Gobain PerformanceAmerica Plastics) Masking tape, ¼inch and½inch Surgical pads 29-Gneedle with Syringe Three-position pad heating bore airtowarm themagnet tubingPlastic (1.5-inch Deluxe Vacuum Hose; Poolstyle) for delivering air for warming Corporation) Milwaukee heavy three-stage duty heatgun(Milwaukee Tool Electric Test fishingline (5lb) containerPlastic (1qt) Isoflurane (Butler Animal Health Supply) Fig.Supplementary 2 no. (model Drager (DRE 19.1)vaporizer Veterinary Supply; Nose cone (Custom design/construction) Animal tray (Custom design/construction; RG-223 coaxial cable (Amphenol) (Tektronix; Tektronix Model TDS1002, two-channel storage oscilloscope digital (Morris Instruments; Morris no. RFtuning unit—model 505NV Haskris Model WW2 water chiller (Haskris Company) EXPERIMENTAL SETUP Varian/Agilent MRIscanner (Agilent Technology), in asdescribed equipped 40-cm clear-bore magnet (4.7 T; Oxford Instruments; MultiHance (Gd-BOPTA MRcontrast agent; Bracco Imaging) gbody weight,1 mgper described aspreviously 100 mgcc cat. no. U2500), dissolved insterile salineatafinal concentration of Urethane, mouselung cancer Chemical, (Sigma primary for of induction Chemical,(Sigma cat. no. T3924) Trypsin-EDTA solution, adherent cells grown inculture of for liberation streptomycin solution (cat. no. P7539), from all Chemical Sigma 10% (vol/vol)(FCS, serum cat. fetalcalf no. F2442), andpenicillin- Tumor RPMI1640(cat. medium consisting growth of no. R0883), Culture Collection) melanoma cell line(cat.B16 malignant no. CRL-6475, TypeAmerican cancer, from both Jackson Labs age (stock no.of 000646), which are susceptible to urethane-induced lung lung by metastases i.v.of tumor injection, and A/J malemice, 6–8weeks Male C57BL/6mice, age (stock no. 6–8weeks of 000664), for establishment ATER I ALS

−

1 Supplementary Fig.Supplementary 2 , (i.p.) andinjectedintraperitoneally for atotal dose of Supplementary Fig.Supplementary 2 )

LCD monitor ) Supplementary Fig.Supplementary 1

+ )

RFSweeper: 1.6–505MHz 1 0 Supplementary Fig. 1

)

) - benzo[ for mouse lung of induction cancer, the use other of carcinogens, such as 100 tion of mg ml To induce tumors, dissolve urethane in sterile saline to a final concentra MRI and 4–5 gross inspection months after carcinogen administration. and is injection about 2 h per group 10 of mice. Tumors become byvisible tumors primary Induction of REAGENT SETUP article. to this necessary MATLAB *.m files are included as below. The described are code this using calculation mass tumor and segmentation lung automated for and code MATLAB this monary tumor burden in the mouse. The procedures for installing time and effort necessary to quantitatively monitor metastatic pul the both decrease to laboratory our in extensively used been has MATLABand within developed and written code using mented lung for segmenting developed acquisition on theMR console. data to used thatcanbe trigger signal transistor-transistor (TTL) logic System, no. model 1025(SAInstruments). purchased and used, such astheSmall Animal Monitoring andGating animal monitoring unitthatincludes monitoring canbe respiratory in Garbow in-houseat designed/constructed Washington University; detailsare available unit monitoring Respiratory chest the of mouse cavity should be selected. coilsmallest into which the mouse can be and inserted that covers the full coil is preferred to a surface coil for this application. As a general rule, the MRI data. the Because of need to image the full the lungs of mouse, a volume Note ‘birdcage-style’ RF coil and shield (Stark Contrast; Fig. 1e Company, (International Electric model GPA-200-350; Technologies (Herley Industries, model amplifiers M3200)); gradient consoleelectronics ( 650 coils ( Agilent 20-cm inner diameter gradient self-shielded, high-performance DirectDrive MRI consoleelectronics ( Varian/Agilent MRI scanner EQUIPMENT SETUP (250 40- 1×10 sterile salineataconcentration of the cells to remove trypsin-EDTA. of traces all Next, resuspend tumor cells in 37 °C for 5 min. Neutralize the trypsin-EDTA with FCS at a 1:1 ratio and wash it reaches 50%confluencetrypsin-EDTAwith 1× by incubating solution at (at~50%confluence, growth phase of logarithmic as B16melanoma show optimal requisite animaleuthanasia ~3weeks after injection. Tumor cell linessuch by 2weeks inspection by aftertumors MRIandvisual visible injection, with injection, 2.5×10 consisting of tumor is1h. cell-line injection In ourexperience, suspension asingle-cell tumors metastatic Intravenous of injection 250 corresponds to 10 other theprogram. versions of MATLAB (7.0.4(R2007a)and7.5.0(R2007b)), likely inmany run andwill Inject mice i.p., 1mgkg atadose of The keyThe requirement isthatthemonitoring unitgenerate a0/+ For tumor injection, B16 melanomalift off from a tissue culture flaskwhen µ µ µ m cell strainer to obtain a single-cell suspension. tom cell asingle-cell obtain strainer Inject 2.5×10 : µ a wide variety of different of a RF variety wide coils can be used for the acquisition of s); high-power RF amplifiers,withinthe DirectDrive installed MRI l of the100mgml l of Supplementary Fig.Supplementary 1d l) intravenously metastases. to establish pulmonary uniform ); Dell Linux PC α ]pyrene, has also been described et al. et 5 0 (

− µ Supplementary Fig.Supplementary 2b

1 . urethane (Although represents our laboratory’s choice l per g of body weight (i.e., body weight gof a25-gmouseisinjected with l per Supplementary Fig.Supplementary 1c

−

LCD monitor (Dell, model no. 390); and a 2.5-cm urethane solution). natureprotocols

; strength: max gradient 28 G cm

The The scanner setup includes: a Varian/Agilent Our respiratory monitoring unitwas Our respiratory 5 The The total time for carcinogen preparation B16mousemelanoma tumor cells, yield will in vivo 4 Supplementary Fig.Supplementary 1b 9

− . This new algorithm was imple was . algorithm new This growth when injected during the injectedduring when growth

1 , usinga100mgml 6 4 cells mlandfilter per through a 5 .) ). Alternatively, acommercial ; 2× American Microwave

The totalThe required time for

| Supplementary Fig.Supplementary 2a in vitro VOL.7 NO.1VOL.7 Supplementary Supplementary Data Supplementary Supplementary protocol ).

−

1 , solution; this c

− );

1 , rise time: | Magnex/ 2012

5 V 5

cells

131

- -

© 2012 Nature America, Inc. All rights reserved. proper positioning of the animal withinthe magnet. (THK) (TE) parameters: repetition time (TR) view along the midline of the mouse). Use the following 15| images and calibrate the RFcoil. 14| from aheat gunthrough the bore of the magnet. throughout the imaging experiments bycirculating warmair 13| universal instructions canbeprovided). tuning are highly coildependent, and thus no simplesetof frequency. Adjust the coiltuning asnecessary (the details of unit, observing reflected poweratthe hydrogen resonance 12| field/gradients ( 11| 10| 9| 8| animal should beplacedsuch thatitschest cavityisatthe center of the coil,asshown in 7| 6| 5| 1-ml syringe witha29-Gneedle. 4| ? with masking tape( snout into the cone, loopthe line around the tubing usedtodeliver the anesthesia (tokeep itfrom slipping) and secure it nose cone withastrand of fishing line loopedaround itsupperfront teeth.Gently pullthe fishing line to draw the animal’s 3| ? DirectDrive console. Fig. 2b 2| ? anesthesia withisoflurane, 2–3%(vol/vol)inO 1| M PROCE 132 protocol

R TROU TROU TROU

dataacquisition

Collectagradient-echo scout scan(single slice, sagittal Maintain the mouse’s body temperature at37±1°C Check the tuning of the RFcoil,using the Morris tuning Position the animal tray inthe magnet, being sure toplacethe RFcoilassemblyatthe center of the magnet Placethe fullassemblyinto the animal tray ( | Calibrate andplanmouselungimages. VOL.7 NO.1VOL.7 Insert the RFcoilinto the RFshield ( Position the RFcoilassembly. Position the mouse prone withinthe RFcoil.The details of securing the mouse are highly coildependent, butthe Stretch the mouse’s hind legsand tapethem toits tailusing ½-inch masking tape. While maintaining anesthesia withisoflurane/O Route the output of the isoflurane vaporizer through the home-built respiratory monitoring/gating unit ( Position themouse. Deliver contrast. Anesthetize themouse.

D 5ms, field of view (FOV) = ). Attach the TTLoutputof the gating unittothe External InputBNCconnector onthe backof the Varian B B B

URE 1mmand data matrix LES LES LES H H H OOT OOT OOT | 2012 Fig. 3 I I I N N N Fig. 3 Delivercontrast byinjecting a500- G G G |

● natureprotocols e ).

Position the mouse within the RFcoil. T IMI a ). To anesthetize the mouse, placeitinto aplastic container attached tothe vaporizerand induce N G

40minpermouse =

= 128×128,toverifythe

3×cm

50ms, echo time Fig. 3 Collectscout 2 , slice thickness 2 c . ). Fig. 3 2 (1–1.5%,vol/vol;O

µ ).

l bolusof MultiHance (Gd-basedMRcontrast agent) i.p.using a ( inside the RF shield; ( coil ( ( Figure 3 c a e a ) ) anesthetized mouse placed inside the nose cone and then inside the RF ) ) animal tray placed inside the magnet. RF coil RF shield b ); ); (

| c Anesthesia and RF coil/animal tray setup for MR acquisition: ) ) RF coil assembly, including an anesthetized mouse, placed 2 flow of 50–100ccmin d ) ) entire coil assembly, placed within the animal tray; d Animal tray assembly RF coilandshield Figure 3

−

1 ), secure the mouse tothe b b e . Animal trayassembly Mouse insideRFcoil S upplementary

© 2012 Nature America, Inc. All rights reserved. custom-written respiratory-gated spin-echo pulsesequence. 20| ? necessary, to permit respiratory monitoring/gating with a19| variable number of image slices collected per respiratory period ( within the scanner’s operating software. the intensity of the watersignal asafunction of pulselength. Enter the appropriate valuesinto the pulsecalibration table 18| of 100Hzorlessaftershimming. to improve itshomogeneity. At 4.7T, wetypically observe aproton line width (full-width athalf-maximum) for water 17| ? animal tray, as necessary, sothatthe mouse’s spine isatthe topof the transaxial image. center of the chest cavity)withthe same experimental parameters usedtocollectthe sagittalimage (Step15).Rotatethe 16| ? the sagittalimage. Slide the animal tray in oroutof the magnet, asnecessary, sothatthe mouse’s chest cavity ispositioned atthe center of ?

B shuffling of slices. standard practice when acquiring non–respiratory-gated, spin-echo data. Asimplemacro canbewrittentoaccomplishthe necessary PE period, thiscanleadtosaturation and subsequent diminution of signal. Thisshuffling isthe equivalent of ‘interleaving’ slices, a ‘bleeding’ of the RFexcitation into adjacent slices. As data from slices withinagrouping are collectedwithinthe short time of asingle acquired during the same PEperiod. The frequency-selective pulsesusedtoselectindividual slices are not ideal and there canbesome 6. Finally, the order inwhich data for the slices iscollectedmust be‘shuffled’ toensure thatimages from adjacent slices are not this limitation when designing respiratory-gated experiments and choosing specificpulse-sequence parameters. following: eprg 5. Withinthisrequirement, the valuesfor totalnumber of slices ( of data ( data collection, the time required toacquire data for one image slice is~50ms. Thus, one cancollectamaximum of perhapsfiveslices anesthesia. Thisrespiratory rate produces aPEperiod of 250–300ms perbreath. Withatypical TEof 30ms and allowing several ms for 4. Letusconsider aspecific example: assume thatthe mouse’s respiratory rate is2s number of slices must beevenlydivisibleby required tocoverallof the slices. Thisscheme placesmodest limitations onthe number of slices thatmust becollected(e.g., the total number of slices inagrouping bythe variable The inner loopcontrols the number of slices (agrouping) tobecollectedduring one PEperiod. Inourpulsesequences, wedefine the 3. To accomplishthis, itisnecessary tobreak upthe loopinthe pulsesequence usedtocollectdata across slices into apairof loops. during the time betweenPEperiods. data, inwhich one wouldlike tocollectdata from several imaging slices during eachpostexpiratory (PE)period, butgather no data to becollected, and acquire data onanindividual slice every(TR/ spin-echo pulsesequences divide the time period defined bythe TRinto 2. Nevertheless, evenpulsesequences thatcontain aflag for triggering acquisition must be modified, as detailedbelow. Standard beginning of the pulsesequence. input signal thatcanbeusedtotrigger data acquisition. Inthe unlikely event thatthisflagis not present, itshould beadded atthe In general, most pulsesequences (including allof those supplied withthe Varian DirectDrive console) contain aflag for an external 1. The spin-echo pulsesequence usedtocollectmouse lung MRimages must bemodified for collection of respiratory-gated data. parameters forrespiratory-gated mouselungimaging TROU TROU TROU TROU

ox Planaseries of coronal slices tocoverthe lungs, typically 21–250.5-mm-thick slices; transfer the plantoa Perform apulsecalibration using aone-pulse spectroscopy pulsesequence; e.g., atfixed transmitter power, measure Using aone-pulse spectroscopy pulsesequence, shimthe magnetic field manually orwithanautomated shimprocedure Use the sagittalimage toplanasecond scoutscan;collectthisscan(single slice, transaxial view through the Collect high-resolution, respiratory-gated spin . Thus, ifone needs tocollect~25slices of data tocoverthe entire lungs, acceptablecombinations of ns and 1 B B B B LES LES LES LES eprg | ns M H H H H

OOT OOT OOT OOT = =

25; 5)perPEperiod (250ms/50 ms odifying thespin-echopulse sequenceandchoosingexperimental I I I I N N N N eprg G G G G

5;ns

24; eprg

= eprg

4; eprg

). = ns

- 5).Clearly, data for asmaller number of slices canalsobeacquired perPEperiod. . The outer(existent) loopthen controls the number of such groupings thatare echo images of the mouse lung.

28; eprg

= ns

4.Although not seriously constraining, the usermust beaware of N ) seconds. However, thisisnot appropriate for respiratory-gated ) and N eprg equal parts, where must beselectedsothat

− Modify the spin-echo pulse sequence, as

1 , anormal rate for awarmanimal under proper

N natureprotocols isthe number of slices of image data ns isaninteger-multiple of

| eprg VOL.7 NO.1VOL.7 include the protocol

| 2012 Box

in vivo

1 133 ).

© 2012 Nature America, Inc. All rights reserved. entry/label withinthe ROI manager corresponding tothe outlined segment of the image. 36| an ROIdrawn around one tumor inasingle image slice. vertex tothe point of origin for the ROI)orbyleft-clicking asecond time onthe point of origin of the ROI. completely outlined. Closethe ROIbydouble-clicking atany point (ImageJ willautomatically connect the ‘double-clicked’ 35| polygon selection drawing tool(third icon from the left)onthe ImageJ menu ( 34| is sections, in coronal later 24 shown containing montage, full A montage. typical a create to used images 20–24 the of ROIs, out defining four of method the Toillustrate images. of montage the overlap not does that screen the on 33| traced, into across-sectional area, measured inunitsof mm 32| 31| ‘Known distance’ (asdescribed inStep22,ourdefault setting is25mm). 30| pixels’ (asdescribed inStep22,ourdefault setting is128). 29| Analyze menu, select‘SetScale’. 28| values withinthe Make Montage dialog box. 27| shown inref. ment of tumor volumes 26| 25| Manual segmentation/volume calculation of primary (discrete) lung tumors ? blurring due tomotion. 24| ? 23| ? collection (e.g., onourscanner, GAIN TE 22| ? 21| 134 protocol

TROU TROU TROU TROU

Add the ROItothe ROI Manager byclicking ‘Add’ withinthe ROIManager dialog box ( Outline the ROIthrough aseries of leftmouse clicks (eachleftmouse click specifies avertex) until the tumor is Create ROIs inthe images thatdefine atumor of interest (Steps34–37;seealso Click OKorhit Inthe SetScaledialog box, enter ‘mm’into the field labeled‘Unit of length’. Inthe SetScaledialog box, enter the length dimension from the FOVusedinacquiring the images into the fieldlabeled Inthe SetScaledialog box, enter the number of pixels inthe acquisition matrix into the fieldlabeled‘Distance in Setthe distance scaletoproperly convertpixels onthe screen into measurable distances (Steps28–32).Under the Byusing ImageJ, create amontage of these images onthe screen: Image We haveshown previously thatcoronal MRimages provide usefulvisualization of lung tumors and facilitate measure Collectfour single-scan data sets. Examine the images, and average them together, excluding scans thatshow obvious Gather respiratory-gated spin-echo MRIdata, withimages collectedduring postexpiratory periods. Collectdata withthe following parameters: TRof ~3s(the exact valuewillbeestablished bythe breathing rate), Shuffle the slice order toensure that data from adjacent slices are not collected during the same breath ( Open the ROI manager: Analyze manager: ROI the Open | Manually segmentprimarylungtumors

VOL.7 NO.1VOL.7 30ms, FOV B B B B LES LES LES LES H H H H 4 OOT OOT OOT OOT | 2 2012 Figure Figure . Use ImageJ toloadastackof contiguous, coronal orientation MRimages thatspanthe entire mouse.

= I I I I

N N N N 25×mm

< G G G G |

Enter natureprotocols 42,4 5 . 3 . The quantitative correlation of MRI-derived tumor volumes withthose measured byhistologyis

. ImageJ willcapture the information needed toconvertsquare pixels, defined inthe ROIs tobe 2 , THK →

= = Tools

48). 0.5mm,data matrix . Draw ROIs around alldiscrete tumors withinthe lung. OpenImageJ software. → ROI Manager. Move the ROI Manager dialog box to a convenient location location convenient a to box dialog Manager ROI the Move Manager. ROI 2 .

128×and scanner gain(GAIN) adjustedfor proper data → Fig. 4 Stacks ●

T IMI S a upplementary Video1 → ). N Make Montage. Accept the default G Fig. 4 15 min per mouse c ). Thisaction willcreate an Figure Figure Figure 4 ). Selectthe Box 1 4 ). shows shows

shows

- © 2012 Nature America, Inc. All rights reserved. 48| 47| 46| 45| 44| 43| 42| 41| 40| 39| click onthe ROItobemodified. 38| its ownuniqueidentification inthe ROI manager. Continue until the ROIs defining the completetumor havebeen drawn onaseries of sequential images. EachROIwillhave 37| tool, a discrete lung tumor is encircled through a series of left mouse clicks. ( for segmentation of discrete lung tumors. ( four segments and the ROI manager. In our experience, this organization of the screen is optimal Figure 4 intensity values. As described in the text, these values are subsequently used to estimate the tumor burden. shown in have been added to the ROI Manager ( image slice and is added to the ROI manager by pressing the ‘Add’ button. ( encircled ROI (e.g., 0085–0135). ( button in the ROI manager creates an entry within the ROI manager uniquely associated with the a a Liver Polygon tool Navigate to the desired folder, enter a name for the set of ROIs under ‘File name:’ with the file extension .zip and click ‘Save’. Click onMore The ROIwillbesavedwiththe file extension .roi. Navigate tothe desired Folder, enter aname for the ROIunder ‘Filename:’ withthe file extension .roi, and click ‘Save’. Click onMore Savethe editedROIbyclicking the Update buttoninthe ROIManager dialog box. To reposition anindividual vertex of the ROI,placethe cursorontopof avertex and drag ittoanew position. To reposition the entire ROI,placethe cursorinside the ROIand drag ittoanew position. Proceed tothe next (contiguous) image containing the same tumor of interest and repeat Steps25–27( consecutive sections Same tumorinfour Edit ROIs—individualROIscanbeeasilyeditedatanytimewithinImageJ(Steps38–41). Save asetof ROIs(Steps46–48). Save individualROIs(Steps42–45).

| f Manual segmentation of discrete tumors. ( . Data in Results include the area of the ROI, the mean intensity of the image within the ROI, as well as minimum (Min) and maximum (Max) → → Save. Save. b d ) A new ROI is created by outlining the tumor on a contiguous e ), selecting the ‘Measure’ button produces the Results screen b b ) As described in the text, after selecting the polygon Ensure thatno individual ROIs are selectedinthe ROIManager dialog box. Click onthe ROItobesavedinthe ROIManager dialog box. a ) Screenshot of a montage containing c e,f c ) After all the ROIs ) Clicking the ‘Add’ c

e in in Manager,the added to ROI as sequentially ( ROI. a as single outlined image are each in and right lungs the left ( tumors. Figure in described as and displayed assembled ( tumors. metastatic 5 Figure natureprotocols Figure Figure 4 Inthe ROIManager dialog box, for segmentation of individual discrete of discrete individual for segmentation b | 4 Manual segmentation of diffuse of diffuse Manual segmentation ) The ROI manager is opened and opened manager is ) The ROI . d f a ) ) The montage is

| VOL.7 NO.1VOL.7 c ) The lung ROIs are ) ROIs The lung Fig. 4 protocol d | 2012 ).

135 © 2012 Nature America, Inc. All rights reserved. tumor progression ortherapeutic response) ( to determine lung tumor burden inthe testseries of mice, perform the stepsinoption D. Except asnoted, the stepsinoption Cshould beapplied toeach(control and tumor-bearing) mouse. To usethisinformation healthy (not tumor-bearing), age-matched control mice ( of mouse, lung images of anappropriate setof healthy control mice must befirstcollectedand analyzed. mice usedtoestablishbaseline image intensity for healthy lung. Ifone wishes tocharacterize tumors inadifferent species described inoption B. Thisprocedure islimitedtostudying mice thatare age and genotype matched tothe cohort of control a series of healthy (i.e., not tumor-bearing), age-matched control mice ( tumor burden. At eachtime point, perform the stepsdescribed inoption A. below). Similarly, decrease (orstability)inaverage, normalized lung intensity indicates regression (orstability)of pulmonary average lung image intensity, compared with(normalized to)thatof liverwithinthe same mouse (Steps54A(v) and 54A(vi) dark, whereas tumor tissueisbright. Thus, asdescribed below, progressive increase in tumor burden leadstoanincrease in application takes advantage of the previously discussedobservation thatinMRimages of lung, healthy parenchyma appears burden across aseries of mice; oroptions Cand Dcanmeasure absolutelung tumor burden across aseries of mice. points (e.g., monitor tumor progression ortherapeutic response); option Bcanbeusedtomeasure relative lung tumor as described below:option Acanbeusedtoassess, qualitatively, lung tumor burden inthe same mouse attwoormore time 54| Manual segmentation/volume calculation of metastatic (diffuse) lung tumors 53| derived volume of asingle tumor, inunitsof mm the images. As described aboveinStep22,ourpreferred slice thickness is0.5mm.The resulting valuewillbethe MRI- 52| divide the areas of the firstand lastROIby2before summing. 51| the columnlabeled‘Area’. The unitswillbespecifiedby‘SetScale’(Steps28–32; 50| in the ROIManager ( from the selectedtumor, then CTRL-leftclick allother ROIs associated withthissame tumor, sothatthey are allhighlighted 49| 136 protocol A B areas across allslices. edge of the tumor. Thus, for ‘edge’ slices, wedivide the ROI-derived cross-sectional area, asdescribed above, by2before summing the the slice. Incalculating tumor volumes, wemake the assumption thatthe tumor is‘halfin’and ‘halfout’of the imaging slices atthe at the edge of atumor, there isno waytoknow, apriori, whether the tumor isfullyorpartially (and towhatextent) contained within knowing. Partial volume averaging results when animaging feature isonlypartially withinthe MRimaging slice. Foranimaging slice plane. Observedimaging features may becompletelyorpartially contained withinthatslice; from the image itself,there isno wayof Each slice withinanimage represents the two-dimensional projection of allfeatures withinthatanatomical slice onto the imaging (iii) (iv)

(ii) To measure absolutelung tumor burden across aseries of mice, youfirst need tocollectimaging data for both aseries of The measurement of relative lung tumor burden across aseries of mice requires thatimaging data firstbecollected for Qualitative assessment of the lung tumor burden should bedone inthe same mouse attwoormore time points. This ) Qualitative assessmentof thelungtumorburden inthesamemouseattwo ormore timepoints(e.g., monitor (i)

ox RepeatSteps34–37and 49–52for eachindividual mouse tumor inorder toestimate itsvolume. Calculatethe tumor volume bymultiplying the summed area of allof the ROIs bythe slice thickness usedinacquiring Sumthe areas of allof the ROIs associated withagiventumor. To properly account for partial volume effects( Inthe ROIManager dialog box, click ‘Measure’. AResultsbox willappearlisting inthe area of eachindividual ROIs under | Estimate tumorvolumefromtheROIsdrawninSteps34–37. Manually segmentmetastatictumors VOL.7 NO.1VOL.7 montage of these images, asdetailed inStep27 ( that allexperimental parameters, including scanner gain,are the same ateachtime point. Set the distance scaletoproperly convert pixels onthe screen into measurable distances, asdescribed inSteps28–32. Open the ImageJ ROIManager asdescribed inStep 33. By using ImageJ, loadastackof contiguous, MRimages (coronal orientation) thatspanthe entire mouse and create a Collect respiratory-gated, spin-echo images of the lung (coronal orientation), asdetailed inSteps 19–24.Becertain 2 | P | artial volume averaging 2012 | Fig. 4 natureprotocols e ). . Tumor burden canbeassessed atthree different levelsof detail and quantification, 3 . Fig. 5 n

5)and atraining series of tumor-bearing mice ( a ). Inthe ROIManager dialog box, leftmouse click the firstROI n

5).Foreachcontrol mouse, perform the steps Fig. 4 ●

T IMI f ). N G ~30 min per mouse n

10). Box

© 2012 Nature America, Inc. All rights reserved. (B) Measurement of relative lungtumorburden across aseriesof mice B B ? An example of such acalibration curveisshown in The resulting plotservesasacalibration curve thatcanbeusedtoconvertMRIlung image intensity into absolutetumor burden. 5. Fita(linear, leastsquares) straight line through allof the points plottedin 4. Plotthe mouse’s total MRI-measured tumor burden versusitscorrected lung weight, asmeasured inStep54C(v). tumor burden. 3. Multiply the mouse’s corrected average, normalized image intensity byitsmeasured lung volume togenerate the totalMRI-measured from the mouse’s average image intensity, toobtainacorrected average, normalized image intensity. 2. Subtract the average, normalized image intensity for healthy lung, computedfrom the cohort of control animals (Step54C(vii)), thickness. 1. Computethe volume of the segmented lungs bysumming the lung ROIareas for everyimage slice and multiplying bythe slice For eachmouse: intensity. Divide the average lung image intensity, calculatedabove, bythe liverimage intensity todetermine anaverage, normalized lung image 5. Computethe average liverimage intensity and normalize the lung image intensity basedonliver: for eachslice, summing these values, and dividing bythe totalof the ROIareas summed overallthe slices. 4. Determine the average lung image intensity across allof the selectedimages bycomputing the product (ROIarea ×mean intensity) 3. Savethe ‘Measure’ table, ifdesired, byselecting ‘SaveAs’ from the ‘File’menu. of mm 2. Aftersegmenting allof the slices containing visiblelung and saving the corresponding ROIs, click ‘Measure’. The ROIarea, inunits heart. to exclude boththe thorax and the spine from the resulting ROIs. We typically omittwoorthree slices perlung image thatinclude the drawing lung ROIs, becareful toavoid image slices containing the heart (which, ifvisible, ishighly blurred becauseof motion), and in eachimage willbeitsownROI( lung ineachimage, asdescribed inSteps34–37and 1. Foreachimage slice containing observablelung, manually segment the lungs (leftand right) and savethe resulting ROIfor each (iii) (vi) (iv) (vi)

(ii) (v) (v) Computethe average, normalized lung image intensity asdescribed in

(i)

ox ox ● ● ● TROU

arising from pockets of lipid orvasculature, respectively. Click ‘Measure’ inthe ROIManager tocalculatethe average image intensity withinthe liverROI. Draw anROIaround alarge, homogeneous section of liver, being careful toavoid obvious regions of either hyper- orhypointensity Select animage slice withclearlyvisibleliver. exclude boththe thorax and the spine from the resulting ROIs. right) and savethe resulting ROI.Indrawing lung ROIs, becareful toavoid image slices containing the heart, and to that allexperimental parameters, including scanner gain,are the same ateachtime point. time points of alongitudinal, time-course study. relative tumor burden withinthe mouse. Values of relative tumor burden canbecompared between animals and across the cohort of control animals, from itsaverage, normalized image intensity. The resulting valueprovides ameasure of arithmetic average) the valuescomputedinStep54B(iii). tumor burden. intensity; anincreased valueindicates increased tumor burden, whereas adecreased valueindicates decreased As detailed abovein Collect respiratory-gated, spin-echo images of the lung (coronal orientation), asdetailed inSteps19–24.Becertain Tumor burden attwoor more time points canbeassessedbycomparing valuesof average, normalized lung image For eachtumor-bearing mouse, subtract the average, normalized image intensity for healthy lung, computedfrom Having established the average, normalized image intensity for healthy lung, the average, normalized lung image Compute the average, normalized image intensity for healthy lung across allanimals byaveraging together (simple, Compute the average, normalized (basedonliver)image intensity for healthy lung, asdetailed in intensity for eachof the tumor-bearing mice iscomputedasdescribed inStep54A. 3 4 2 S , and the mean intensity for the ROIineachimage slice willbedisplayedinthe pop-uptable. upplementary Video1 B LES | | C C H OOT omputing average, normalizedlungimageintensity omputing absolutelungtumorburden I N G Box demonstrates the manual segmentation of mouse lungs using ImageJ. 3 , for eachimage slice containing observablelung, manually segment the lungs (leftand Fig. 5 b shows segmentation of the leftlung; Figure S upplementary Video1 1 . . At the completion of thisstep,eachindividual lung Box 4 Fig. 5 Box , step4thatpassthrough the origin (0,0). c shows segmentation of the right lung). In 3 . natureprotocols

| Box VOL.7 NO.1VOL.7 3 protocol . | 2012

137 © 2012 Nature America, Inc. All rights reserved. 138 protocol B MATLAB ‘SaveAs’ command. overlaid inred ( 10. The program displaysasoutput,withinaMATLAB figure, a montage of lung images withautomatic segmentation boundaries performs thissegmentation isprovided elsewhere inthis protocol. segmentation (128×128voxel images) onanIntel Core 2Quad, 2.4GHzDellcomputer is~3min.Anoutline of how the program 9. The program displaysa‘Progress Bar’that allowsthe usertofollow the statusof the segmentation. The execution time for a15-slice segmentation routine. outlining liveristhe same asthatdescribed for lung instep6.Doubleclicking withinthe liverROIinitiates the automated lung 8. Afterthe program prompt, outline aregion of liverbelowthe lungs inthe displayedimage slice ( segmented area. Similarly, the lowercenter point should bethe corresponding point closesttothe bottomof the image. segmented slice. The uppercenter point should bethe point closesttothe topof the image thatlies along the median line withinthe should bothlie along the median line of the animal, which may bedistinguishable becauseof small amounts of spine visibleinthe 7. Afterthe program prompt, click onthe upperand lowercenter points of the displayed, segmented lungs ( 6. Afterthe program prompt, outline the exterior boundary of the lungs inthe displayedimage slice ( 5. Detailsaboutthe command options canbefound bytyping the command: both leftand right lung. number of the ventral-most slice withsizablelung area and dorsal isthe slice number of the dorsal-most slice withsizableareas of rotate typing: by line command Window MATLABCommand the in arguments as input directly be can information necessary the Alternatively, 4. of the montage, and ending inthe bottom-right corner. In the displayedmontage, images are tiledventral todorsal, from lefttoright and toptobottom,beginning inthe upper-leftcorner for the rotation flag,and a montage of all of the lung images isdisplayedwhen the userisprompted for the slice-number inputs. 3. To aid inthe inputof these values, asingle lung slice from the specified folder of MRimages isdisplayedwhen the userisprompted 2. Execution of thiscommand willcausethe program toprompt the usertoenter allof the necessary pieces of information: 1. The program canberunwithno inputsorarguments, bytyping the following command into the MATLAB Command Window: Refer to How torunthe published paper A more detailed discussion of the algorithm, and the objectivefunction usedfor the optimization steps, canbefound inourrecently calculated from the manually drawn liverROI. 5. The average lung image intensity iscalculatedfrom allof the segmented slices and isthen scaledbythe average liverintensity, 4. The parameter valuesfor the adjacent slices are similarlypropagated dorsally and ventrally until alllung tissueissegmented. for the twoadjacent slices. 3. The optimizedparameter valuesfor the initial slice are usedasinitial valuesfor determining the locallyoptimal parameter values A locallyoptimal setof parameter valuesiscalculatedusing the Nelder-Mead simplex method. 2. Aparametric model of the lung shapeisdetermined from the initial image and the manual segmentation of thisinitial slice. segmented lungs; and amanually outlined area of liverinasingle slice. dorsal-most) image slices tosegment; amanually outlined single, initial slice of lung; the upperand lowercenter points of the 1. As described below, the automatic lung segmentation process acceptsasmanual userinput:the firstand last(ventral-most and Description of thealgorithm

ox ● ● ● ● ● ● help Lung_Segmentation_Main [Avg_intensity,Lung_Volume] ● ● ● ● [Avg_intensity,Lung_Volume] | VOL.7 NO.1VOL.7

To reposition anindividual vertex, leftclick the vertex and drag it toanew position. horizontally, rather thanvertically; Detailed instructions onthisinterface are availableinthe MATLAB help file for the ‘roipoly.m’ function. The polygon isfinalized by double-clicking withinthe closedshapeoutlining the lungs. The ROIcanberepositioned ifnecessary. To reposition the entire ROI,leftclick inside the ROIand drag ittoanew position. The polygon canbeclosedbyreclicking the starting vertex, orbydouble-clicking when selecting the final vertex. The lung isencircled withanROIbyleftclicking onaseries of vertices thatcreate anoutline of the desired polygon. The leftand right lungs should not beseparated, butrather should beoutlined withinone continuous boundary. The slice number of the dorsal-most slice withsizableareas of bothleftand right lung ( The slice number of the ventral-most slice withsizablelung area ( A flagindicating whether the images must be rotated 90°before processing (necessary ifthe original images are oriented The directory containing the MRlung images, in*.fdf format, which are tobesegmented ( where hasthe valueof 1or0,depending uponwhether rotation of the image isrequired (1)ornot required (0),ventral isthe slice 5 S upplementary Video2 | directory O | 2012 utline ofthealgorithmforautomatedimagesegmentation 4 Fig. 6 L 9 . ung_ isthe pathof the directory containing the lung images (entered asastring, i.e., withinsingle quotation marks), |

h natureprotocols S ). Thismontage canbesavedfrom withinthe figure, if desired, inavariety of different formats, using the egmenation_Main.m program

= = .

Lung_Segmentation_Main(directory,rotate, ventral,dorsal) Lung_Segmentation_Main Fig. 6 b ) Fig. 6 c ) Fig. 6 Fig. 6 c ) Fig. 6 a Fig. 6 ) g ). The procedure for d Fig. 6e, ). f ). These points (continued)

© 2012 Nature America, Inc. All rights reserved. file, isprovided in of how the program performs lung segmentation, together withinformation onhow torunthe ‘Lung_Segmentation_Main.m’ 57| which the attached MATLAB*.m fileswere saved(Program Files/Matlab/R2008a/toolbox/lung_segment). calculation, asdescribed in 56| lung_segment. single new folder named lung_segment withinthe MATLAB toolbox folder: Program Files 55| A (D) Determinationof absolutetumorburden inatesttumor-bearingmouse ( (viii) C (vii) (vii) utomated utomated segmentation/volume calculation of metastatic (diffuse) lung tumors ? automatic and manual segmentation willnot bethe same (automatic segmentation does not exclude slices containing heart). mixed. Allcontrol, training and testanimal data setsmust besegmented automatically, asthe average lung image intensities basedon 12. Note thatincomputing lung tumor burden, manual (Step54A)and automatic (Steps55–57)lung segmentations should not be Step 22,ourdefault setting is~0.019mm Fig. 6 intensity 11. The program alsooutputs(inthe MATLAB Workspace window) valuesof the liver-corrected lung intensity (variable name (iii) Calculatethe totallung volume, asdescribed aboveinStep54C(viii). (iii) Remove and weigh the lungs. (vi) (vi) (iv) (vi) (iv) ) Measurement of absolutelungtumorburden across aseriesof mice B (ii) Calculatethe average, normalized lung-image intensity asdescribed in (ii) (v) (v)

(i) (i) TROU Runthe file‘Lung_Segmentation_Main.m’, which iscontained inthe supplied package of MATLAB *.mfiles. An outline ox Process theMRimagesinMATLAB. Install MATLAB *.m files. ? that allexperimental parameters, including scanner gain,are the same ateachtime point. that allexperimental parameters, including scanner gain,are the same ateachtime point. ? tumor-bearing mice must befirstcollectedand analyzed different species of mouse, lung images of anappropriate cohort of healthy control mice and atraining series of bearing mice usedtoestablishthe calibration curveof Step54C(viii).Ifone wishes tocharacterize tumors ina studying mice thatare age and genotype matched tothe cohort of control mice and the training setof tumor- together (simple, arithmetic average) the valuescomputedinStep54C(vi). (Step 54C(vii))toyield acorrected average, normalized lung intensity. Subtract the average, normalized lung intensity calculatedfor the cohort of control (i.e., not tumor-bearing) mice Collect respiratory-gated, spin-echo images of the lung (coronal orientation), asdetailed inSteps19–24.Becertain For eachtumor-bearing mouse, computethe absolute tumor burden asdescribed in For the cohort of control mice, computethe average, normalized image intensity for healthy lung byaveraging Compute the average, normalized lung-image intensity asdescribed in For eachtumor-bearing mouse, subtract the average weight for healthy lung, computedfrom the cohort of control For the entire cohort of control mice, computethe average lung weight (simple, arithmetic average). Euthanize the mouse immediately afterthe imaging experiments. Collect respiratory-gated, spin-echo images of the lung (coronal orientation), asdetailed inSteps19–24.Becertain As noted abovefor the measurement of relative lung tumor burden across aseries of mice, thisprocedure islimitedto Multiply the corrected average, normalized lung intensity (Step54D(iv))and the totallung volume (Step54D(iii))to Use the calibration curve generated in Step 54C(viii) to convert this MRI-measured value into absolute tumor tumor absolute into value MRI-measured this convert to 54C(viii) Step in generated curve calibration the Use burden (mg). burden generate the totalMRI-measured tumor burden. animals (Step54C(iv))from itsmeasured lung weight, toobtainacorrected tumor weight. i

). Thisvolume canbeconvertedtophysical units(e.g., mm 5 TROU TROU B ), averaged across allof the segmented slices, and the totalsegmented lung volume invoxels (variable name LES | H C B B OOT LES LES ontinued I N Box H H G OOT OOT 5 I I (seealso N N G G Box Downloadallof the MATLAB*.m filesfrom the

5 . OpenMATLAB (version 7.6.0R2008a)and selectasyourcurrent directory the folder into Fig.

S 3 6 upplementary Video2 ). ). 3 ) bymultiplying the voxel volume bythe voxel size(asdescribed in 4 shows the stepsinthe automated segmentation/volume 9 .

S upplementary Data Box Box 3 3 . . ● natureprotocols →

T Matlab Box IMI N 4 G . →

10 min per mouse and savethem into a R2008a

| VOL.7 NO.1VOL.7 →

= toolbox

protocol Lung_Volume

= |

2012

Avg_ →

; 139 © 2012 Nature America, Inc. All rights reserved. T Troubleshooting advice canbefound in ? MATLAB Workspace window. lung volume ( intensity ( ( the program and displayed as a montage plot. of the liver. ( the lung, the user is prompted to outline a part and identifying the upper and lower points of ( displayed segmented lungs are identified. the upper ( ( lungs in one representative slice is outlined. lung fields, the exterior boundary of the automatic segmentation of all the displayed 1–4, 23 and 24, as illustrated). ( of spine or anterior chest wall (e.g., slice nos. contain lung tissue and some will contain images This step is necessary, as not all images will of the lung are entered into the program. containing the most ventral and dorsal portion enter ‘0’. ( vertical orientation by entering ‘1’, otherwise oriented horizontally are rotated 90° to a be segmented is input. ( Main.m’, the directory containing the files to and running the file named ‘Lung_Segmentation_ metastatic tumors. ( Figure 6 140 i g e,f a 16 15 1–3 S 19–24 protocol ) Values of the liver-corrected, average lung

) After outlining the boundary of the lungs TROU tep b ) Following the automated program prompt,

le | VOL.7 NO.1VOL.7 2

| B c | Automated segmentation of diffuse Avg_intensity ) The numbers of the image slices e LES

Troubleshooting table. ) and lower ( h Lung_Volume ) The lungs are segmented by H Rapid respiration rate in image Mouse spine not vertical Lungs not in image FOV Uneven respiration P roblem OOT | 2012 a ) After opening MATLAB I ) and total segmented N b f ) If necessary, images G | ) center points of the ) are output to the natureprotocols d ) To initiate

T mouse Improper positioning of the induction Oversedation with anesthetic at R Light anesthesia in RF coil The mouse is awake and moving mouse Improper positioning of the Overheated Overheated animal able eason d g a 2 .

h b e Check the O Check isoflurane level in vaporizer and fill, if necessary mouse; increase isoflurane Remove animal tray from magnet; reposition the Rotate animal tray to properly align the spine within magnet Check position of coil within animal tray and tray Decrease isoflurane in steps of 0.1% restore isoflurane to 1%; monitor respiration Turn off isoflurane; wait for respiration to recover; S Adjust the temperature/flow of heat gun Increase isoflurane in steps of 0.25% Check the O olution 2 2 flow rate and adjust as necessary tank, replace if empty c f i (continued)

© 2012 Nature America, Inc. All rights reserved. the Barnes-Jewish Hospital Foundation, and the generous support of support generous the and Foundation, Hospital Barnes-Jewish the Grant; Research Foundation Society/Lungevity Thoracic American an CA131097); (KO8 Grant Institute Cancer Center, NIH/National Cancer Siteman the from Grant Research Internal Society Cancer American CA91842); (P30 Center Cancer Comprehensive NCI an Louis, St. in University Washington at Center Cancer Siteman J. Alvin the CA83060); (U24 Program Resource Imaging Animal Small Institute Cancer NIH/National the CCF-0963742; Grant Foundation Science National Study; Graduate Women for in T. Fellowship Olin A is availableNote: information the via HTML of version Supplementary this article. training orbackground inMRIdata acquisition orimage analysis before working inourlaboratory. are routinely executed inourlaboratory byaveterinary technologist withbasic animal handling skills, butno previous (data acquisition) and 5–10 animals (image analysis) for any individual withabasic science background. These procedures On the basisof ourexperience withthese procedures overthe past5years, weanticipate alearning curveof 24animals ANT Steps 55–57,Automated segmentation/volume calculation of metastatic (diffuse)lung tumors: 10minpermouse Step 54,Manual segmentation/volume calculation of metastatic (diffuse)lung tumors: ~30minpermouse Steps 25–53,Manual segmentation/volume calculation of primary (discrete) lung tumors: 15minpermouse Steps 1–24,MRdata acquisition: 40minpermouse ● T Sheldon and Charlotte Rudnick. Charlotte and Sheldon a Box 5 and 54C, 54D S ckno

tep b T le IMI I Box 4 C w 2 I le PATE N | dgm G

Troubleshooting table (continued). D ents segmentations Poor-quality automatic Segmentation_Main’’ variable ‘Lung_ ‘Undefined function or MATLAB error message calibration curve Poor correlation of Apparent loss of respiration Depressed respiration rate P roblem RESULTS

This work was supported by a Mr. and Mrs. Spencer a Mr.by Spencer Mrs. and supported was work This

parameters Improper selection of imaging Improper positioning of the mouse Improper manual initialization proper directory *.m files not installed in the for intensity normalization Improper selection of liver regions Inconsistent imaging parameters nose cone The mouse is separated from the pressure Excessive oxygen flow/back The mouse is cold R eason

com/reprints/index.html at online available is information permissions and Reprints Published online at interests. CO use MRI for monitoring primary mouse lung tumors. provided lung tumor-bearing animals and were instrumental in initial efforts to A.N. helped with data analysis; V.V.A. provided technical support; M.Y. and H.G.V. collected all the MR images; A.E.G. and D.K. helped with experimental design; experiments, performed the data analysis and wrote the manuscript; J.A.E. AUT

M H PET OR I

N CONTR G G FI Check manual inputs, including the starting and ending directory specified in Step 55 Verify that the provided files are installed into the homogeneous portion of the liver, contains only liver and is relatively Check that the liver region selected contains a large control and tumor-bearing animals Check that imaging parameters are identical for all Remove animal tray from magnet; reattach the nose cone Reduce the oxygen flow/pressure Adjust the temperature/flow of heat gun S Reimage if necessary good contrast between healthy lung tissue and tumor. Check that the images are sharp and low in noise, with spine being positioned at the top of the transaxial image the dorsal-most images. This corresponds to the mouse’s Check that the mouse’s lungs are roughly symmetric in with the instructions provided in selection of the center points, to verify that they comply slice numbers, the manual lung segmentation and the olution NANC IB UT http://www.natureprotocols.com I I AL ONS I / . NTERESTS A.S.K., A.S.K., V.K.T. and J.R.G. designed the imaging natureprotocols The authors declare no competing financial declare no The financial authors competing Box 5

/ | . VOL.7 NO.1VOL.7

protocol http://www.nature. | 2012

level allogeneic chimerism achieved by prenatal tolerance induction andinduction tolerance prenatal by achieved chimerism allogeneic level Radiopharm. Biother. Cancer tumor.murine in radiotherapy to response cancer lung Lewis the (2010). 5679–5685 irradiation. local after regrowth tumor facilitates cells research. cancer animal small for tool emerging an tomography: computed X-ray High-resolution (2010). using mouse the design. scanning. CT imaging.bioluminescence andCT, small-animalPET, small-animal combined with mice in metastases Res. Cancer microenvironment.lung the in cells tumor of growth not but survival to contributes cells marrow–derived bone from metalloproteinase-9 Matrix mice.transgenic in cancer biology.radiation bar.the TBK1. require cancers KRAS-driven tolerancemouseain model oflung cancer. expressiontumor-associatedofa antigen reveals multiple levels ofT-cell Protoc. Nat. recombinase.Cre of delivery lentiviral or adenoviral using models mice. in cancer lung onset (2011). 72–85 progression. tumormalignant adenocarcinomasdelay lung progression. cancer lung cancer.lung of models imaging.tomography microcomputed genomeassociation analyses ininbred mice. exposure.mucosal cancer.in surveillance origins. similar of tumors rejecting versus growing progressively in cells T effector and regulatory of autoimmunedepigmentation. inducesrejection ofsubcutaneous andmetastatic tumors accompanied granulocyte/macrophageby colony-stimulating factor (GM-CSF)-producingmelanoma using anti-cytotoxic vaccines lymphocyte-associatedT antigen (CTLA-4)4 and melanoma.metastatic patients. cancer vaccinated previously in 4 antigenlymphocyte-associated T cytotoxic (2006). 221–232 Clin. J. Cancer carcinoma. colorectal with patient a in metastasis lung a Peranteau, W.H., Hayashi, S., Hsieh, M., Shaaban, A.F. & Flake, A.W.High- Flake, & A.F. Shaaban, M., Hsieh, S., W.H.,Hayashi,Peranteau, enhances oligodeoxynucleotide1826 W.CpG Chen, & T.Qiao, S., Yuan, S.V.Kozin, D.K. Johnson,Hunsicker, P.R.& S.J., Kennel, S.S., Gleason, M.J., Paulus, E. Namati, D.R. Aberle, R.J. Klaveren, van C.M. Deroose, L.M. Matrisian, & D.L. Gorden, B., Fingleton, K.J., Carter,H.B., Acuff, lung Chemopreventionof M. You, & Y.Wang, Z., Zhang, R.A., Lubet, D.G. Kirsch, C.F. Kim, D.A. Barbie, Cheung,A.F., Dupage, M.J., Dong, H.K., Chen,Jacks, &J. T.Regulated cancer lung mouse ConditionalT. Jacks, & A.L. Dooley, DuPage,M., L. Johnson, DuPage,M. D.M. Feldser, N. Guilbaud, H. Fushiki, Liu,P. via down-regulationresponse: immune the and cells T J. Bienenstock, immune- on perspective EpidemiologicO.J. Finn, Cramer,D.W.& analysis Comparative Schreiber,R.D. & C.S. Hsieh, R., Uppaluri, J.D., Bui, vanElsas, A.,Hurwitz, A.A.Allison,& J.P. Combination immunotherapy ofB16 F.S.Hodi, F.S.Hodi, S.P.Leong, 2010. statistics, Cancer E. Ward, & J. Xu, R., Siegel, A., Jemal, of M.W.Expectoration Saif, & N. Suh, D., Cornfeld, M., Davies, C., Ghetie, | VOL.7 NO.1VOL.7

etal.

7 60 Radiology , 283–286 (2008). 283–286 , Cold Spring Harb. Symp. Quant. Biol. Quant. Symp. Harb. Spring Cold et al. et , 277–300 (2010). 277–300 , et al. et al. et

et al. et Candidate lung tumor susceptibility genes identified through whole- et al. et et al. et

et al. et 66 et al. et et al. et 4 et al. et et al. et et al. et et al. et N. Engl. J. Med. J. Engl. N. et al. et , 1064–1072 (2009). 1064–1072 , et al. et , 259–266 (2006). 259–266 , | Mouse models of human non-small-cell lung cancer: raising cancer: lung non-small-cell human of models Mouse 2012 Improved survival with ipilimumab in patients with patients in ipilimumab with survival Improved of blockade antibody of effects clinical and Immunologic Longitudinal assessment of lung cancer progression in progressioncancer lung of assessmentLongitudinal Recruitment of myeloid but not endothelial precursorendothelial not but myeloid of Recruitment Proc. Natl. Acad. Sci. USA Sci. Acad. Natl. Proc. Quantification of mouse pulmonary cancer models by models cancer pulmonary mouse of Quantification Clinical patterns of metastasis.of patternsClinical Endogenous T cell responses to antigens expressed in expressedantigens to responses cell T Endogenous

in vivo in Somatic activation of the of activation Somatic Imaging primary lung cancers in mice to study to mice in cancers lung primary Imaging The National Lung Screening Trial: overview and study and overview Trial:Screening Lung National The Systematic RNA interference reveals that oncogenic that reveals interference RNA Systematic 258 Int. J. Radiat. Oncol. Biol. Phys. Biol. Oncol. Radiat. J. Int. Antitumor activity of S 16020-2 in two orthotopic two in 16020-2 S of activity Antitumor Can. Respir. J. Respir. Can. Stage-specific sensitivity to p53 restoration during restoration p53 to sensitivity Stage-specific et al. et Multimodality imaging of tumor xenografts and xenografts tumor of imagingMultimodality , 243–253 (2010). 243–253 , | N. Engl. J. Med. J. Engl. N. Curr. Opin. Immunol. Opin. Curr.

Anticancer Drugs Anticancer micro-CT imaging.micro-CT natureprotocols Management of lung nodules detected by volume by detected nodules lung Managementof Neoplasia Nature Cancer Res. Cancer J. Nucl. Med. Nucl. J. Chest

Nature J. Exp.J.Med. 26

361 , 203–208 (2011). 203–208 ,

468

125

, 2221–2229 (2009). 2221–2229 , 2 410 (suppl A), 27A–30A (1998). 27A–30A A), (suppl Nature , 62–70 (2000). 62–70 , , 572–575 (2010). 572–575 ,

, 144S–147S (2004). 144S–147S , Cancer Sci. Cancer 66

, 1111–1116 (2001). 1111–1116 ,

363

48 190 8 , 7301–7309 (2006). 7301–7309 ,

CancerRes. Nat.Genet. , 276–282 (1997). 276–282 ,

462 , 295–303 (2007). 295–303 , 105 Med. Phys. Med. , 711–723 (2010). 711–723 ,

, 355–366, (1999). 23

K-ras 70 , 108–112 (2009). 108–112 , , 265–271 (2010). 265–271 , , 3005–3010 (2008). 3005–3010 ,

, 241–250 (2005). 241–250 , 100 Cancer Metastasis Rev.Metastasis Cancer oncogene causes early causes oncogene

38 68 , 1544–1549 (2009). 1544–1549 ,

, 973–977 (2010). 973–977 , 37 Cancer Res. Cancer , 888–895, (2006). , 9459–9468, (2008). , 4793–4805 , Clin. Colorectal Clin. Cancer Cell Cancer CA Cancer CA

, 19

, ,

31. 43. 42. 41. 40. 39. 38. 37. 36. 35. 34. 33. 32. 60. 59. 57. 56. 55. 54. 53. 52. 51. 50. 49. 48. 47. 46. 45. 44. 58.

postnatal nonmyeloablative bone marrow transplantation.marrow bone nonmyeloablative postnatal resonance imaging.resonance magnetic by rodents in adenocarcinomadevelopment of monitoring tumors. liver mice. Neurooncol. J. gliomas.low-grade of therapeuticsexperimental for model a gliomas: rat Spectroscopy Methods and Theory imaging. Ther. Biol. Cancer autoimmunity.antitumor cells. T proliferating cells. T CD4+ requiresproliferation CD8 memory-like protective ofgeneration (2002). 2225–2234 imaging-based high-throughput assay.high-throughput imaging-based USA Sci. Acad. Natl. Proc. PET/CT.using carcinomas lung cell nonsmallkinase-expressing (EGFR) mice. inbred in cells 180 sarcoma proteolyticnovelmolecularenzymeactivityreporter.usinga imaginglungcancer.of USA Sci. Acad. Natl. Proc. xenografts. tumor lung with mice on study micro-PET a erlotinib: [11C]-labeled (2009). 3837–3846 rodents.free-breathingof imagingmicro-CT correlated SSCE-MRI. and DCE- by assessed isoforms VEGF differentoverexpressing cancers lung in angiogenesis (2009). 1507–1514 MRI. contrast-enhanceddynamic T1-weighted Eng.Reson.Magn. B: PartReson.,ConceptsMagn. experiments.MRS andrespiratory MRIactive ingatingor passive devicefor Med. Reson. MRI. via lung mouse in burden tumor of analysis Quantitative herbs. Chinese of mixture a by Phys. studyradiation biology:Kirschregardtoin Res. metastases. lung melanomahuman of growth in variability Carcinogenesis mice. strain A/J in progression tumor lung difluoromethylornithine during 1996). Press, (Futura 401–415 A.) Cutillo, (ed. in lung magneticrodentsresonanceby imaging. 4953–4958 (2000). 4953–4958 Hamilton, S.E., Wolkers, M.C., Schoenberger, S.P. & Jameson, S.C. The S.C. Jameson, Schoenberger,S.P.& M.C., Wolkers, S.E., Hamilton, Garbow, J.R., Zhang, Z. & You, M. Detection of primary lung tumors intumors primarylungDetectionofYou, M. & Garbow, Z.Zhang, J.R., Quantitative M. You, & R.A. Lubet, Y.,Wang, M., Wang, J.R., Garbow, S.R. Cai, Abdulkadir,S.A. P.E.Kish, B. Mann, & J. Akitt, Becker,E. & Farrar,T. Hashemi,Bradley,R.H.& W.G. Jr. PET animal SmallT. Receveur, V.C.& Soon, M.A., Miller, G.D., Hutchins, Y.H.Jeon, Dummer,W. A. Shvets, Zhang, Y.Zhang, H.H. Yeh, Weissleder,R. & Bredow,S. U., Mahmood, C.H., Tung, Grimm,J. C.P.Phenix, A.A. Memon, phase- Kalender,W.A.Respiratory & R.M. Lapp, Y.,Kyriakou, D., Ertel, A. Yuan, J.H. Naish, simple,hardwarerobustA Conradi,M. & S.-K. Song,Garbow, J.,Dugas, J., A. Nehorai, & J.A. Engelbach, A.S., Krupnick, J.R., V.K.,Garbow,Tidwell, Y.Wang, Garbow,Ackerman,& J.R.ImagingJ.J. primary tomice cancerslung in B.M. Carreno, M.W.Anderson, the of microscopy MR G.A. Johnson, & Cofer,G. S., Dewalt, L., Hedlund, Vikis, H.G. Vikis,

15 2010;76:973–977). Nat. Med. Nat. , 3277–3286 (2009). 3277–3286 , Application of Magnetic Resonance to the Study of the Lung the of Study the to Resonance Magnetic of Application et al. et et al. et et al. et ILAR J. ILAR et al. et et al. et et al. et et al. et et al. et et al. et et al. et doi:10.1002/mrm.22951 (27 September 2011). September (27 doi:10.1002/mrm.22951 Cancer Res. Cancer et al. et (Stanley Thomes, 2000). Thomes, (Stanley et al. et et al. et Int. J. Oncol. J. Int. et al. et

et al. et A mouse model for developing treatment for secondary for treatmentdeveloping for model mouse A Functional and structural characteristics of tumor of characteristics structural and Functional Chemoprevention of lung squamous cell carcinoma in mice in carcinoma cell squamous lung Chemopreventionof 29 53 Use of geneofexpressionUse profiling direct to Identification of inhibitors of ABCG2 by a bioluminescence a by ABCG2 of inhibitors Identification of Magnetic resonance imaging of ethyl-nitrosourea-inducedof imagingresonance Magnetic

Molecular imaging of active mutant L858R EGF receptor EGF L858R mutant active of imaging Molecular et al. et Impaired negative regulation of homeostatically of regulation negative Impaired Immune response to firefly luciferase as a naked DNA. a naked as luciferase firefly to response Immune 7

Strain-specific susceptibility for pulmonary metastasis of metastasis pulmonary for susceptibility Strain-specific et al. et

Modeling of contrast agent kinetics in the lung using lung the in kinetics agent contrast of Modeling , 1594–1600 (2008). 1594–1600 , 49 , 243–257 (2001). 243–257 , T cell homeostatic proliferation elicits effective elicits proliferationhomeostatic cell T , 101–107 (2001). 101–107 , 6 Imaging of enzyme replacement therapy using PET.using therapyreplacement enzyme of Imaging Positron emission tomography (PET) imaging with imaging (PET) tomography emission Positron , 781–786 (2007). 781–786 , Immunodeficient mouse strains display marked display strains Immunodeficient mouse (Academic Press, 1971). Press,(Academic Clin. Cancer Res. Cancer Clin. , 54–65 (2008). 54–65 , Blood NMR and Chemistry: An Introduction to Modern NMR Modern to Introduction An Chemistry: and NMR Effect of dietary green tea extract and aerosolized and extract tea green dietary of Effect Impaired prostate tumorigenesis in Egr1-deficient in tumorigenesisprostate Impaired Pulse and Fourier Transform NMR—Introduction to NMR—Introduction Transform Fourier and Pulse

Proc.Natl.Acad.USASci. 69 Int. J. Radiat.Oncol.Phys.Biol.J.Int.

J. Clin. Invest. Clin. J.

, 873–878 (2009). 873–878 , 108 107

113 27 , 113–120 (2005). 113–120 , , 1603–1608 (2011). 1603–1608 , (2011). 10842–10847 , , 622–625 (2009). 622–625 , Cancer Prev. Res.Prev. Cancer MRI: the BasicsMRI:the PLoS One PLoS Cancer Res. Cancer Nat. Immunol. Nat.

14 Cancer Res. Cancer , 1363–1367 (2008). 1363–1367 ,

Cancer Res.Cancer 110 +

T cells during homeostaticduring cells T et al. et 6 , e16062 (2011). e16062 , , 185–192 (2002). 185–192 ,

Magn. Reson. Med. Reson. Magn. 70 (Williams Wilkins,& 1997). (

102 2 , 4859–4867 (2010). 4859–4867 , 21B

Int. J. Radiat.Oncol.Biol.J.Int.

, 634–640 (2009). 634–640 , 7 69

64 , 475–481 (2006). 475–481 , , 14404–14409, (2005). , 40–48 (2004).40–48 , , 5867–5875 (2009). 5867–5875 , , 2740–2742(2004)., In vivo In Phys. Med. Biol. Med. Phys.

79 in vivoin Blood , 959 (2011).959 , Clin. Cancer Clin. Cancer Res.Cancer imaging of imaging Magn. molecular

100 61

60 54 , ,