Pediatric obesity phenotyping by magnetic resonance methods

Wei Shena, Haiying Liub, Mark Punyanityaa, Jun Chena, and Steven B. Heymsfieldba New York Obesity Research Center, St. Luke’s-Roosevelt Hospital, Institute of Human Nutrition, Collegeof Physicians and Surgeons, Columbia University, New York, USA

b Merck & Co., Rahway, New Jersey, USA

AbstractPurpose of review—Accurate measurement of adiposity in obese children is required forcharacterizing the condition’s phenotype, severity, and treatment effects in vivo. Non-invasive andsafe, magnetic resonance imaging and spectroscopy provide an important new approach forcharacterizing key aspects of pediatric obesity. This review focuses on recent advances in non-invasive magnetic resonance imaging and spectroscopy for quantifying total body and regionaladiposity, mapping adipose tissue distribution, and evaluating selected metabolic disturbances inchildren. The aim is to provide an investigator-focused overview of magnetic resonance methods foruse in the study of pediatric body composition and metabolism.

Recent findings—Whole body axial images can be rapidly acquired on most clinical magneticresonance imaging scanners. The images can then be semi-automatically segmented intosubcutaneous, visceral, and intramuscular adipose tissue. Specific pediatric studies of errors relatedto slice gap and number are available. The acquisition of scans in healthy and premature infants isnow feasible with recent technological advances. Spectroscopic, Dixon, and other approaches canbe used to quantify the lipid content of liver, skeletal muscle, and other organs. Protocol selection isbased on factors such as subject age and cost. Particular attention should be directed towardsidentification of landmarks in growth studies. Recent advances promise to reduce the requirementof subjects to remain motionless for relatively long periods.

Summary—Magnetic resonance imaging and spectroscopy are safe, practical, and widely availablemethods for phenotyping adiposity in children that open new opportunities for metabolism andnutritional research.

Keywordsbody composition; genetics; magnetic resonance spectroscopy; nutritional assessment

IntroductionAlthough the increasing prevalence of excess adiposity in children was first reported almosttwo decades ago [1], childhood obesity continues to increase at an alarming rate [2–4]. Obesityin childhood is not only related to health risks [5] such as the metabolic syndrome [6] andinsulin resistance [7], but also often progresses to adult obesity [8,9] associated with morbidityand mortality [10,11]. Accurate measurement of adiposity in phenotyping children is centralto contemporary genetic and clinical research.

Correspondence to Wei Shen, MD, New York Obesity Research Center, 1090, Amsterdam Avenue, 14D, New York, NY 10025, USA,Tel: +1 212 523 1738; fax: +1 212 523 3571; e-mail: ws2003@columbia.eduSponsorship: Supported by National Institutes of Health grant DK-PO1-42618.

NIH Public AccessAuthor ManuscriptCurr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.Published in final edited form as:

Curr Opin Clin Nutr Metab Care. 2005 November ; 8(6): 595–601.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Many adiposity phenotyping methods are currently available that range from simple to complexand that provide estimates of total and regional body composition [12]. Refining phenotypes,improving their resolution in detecting small between-individual differences, is a high researchpriority.

The currently accepted adiposity phenotyping reference methods are magnetic resonanceimaging (MRI) and computed tomography (CT). Both methods provide high-resolution imagesthat are usually then quantified as specific component areas or volumes. MRI is increasing inuse, particularly in children, because of radiation exposure associated with CT. MRI has beenused in studying physiological conditions such as adiposity [13], growth and development[14,15], and pathological conditions such as lipodystrophy [16] and muscular dystrophy [17].

While in adults MRI had been extensively used for body composition studies, adaptation isneeded when it is applied in children. In this review we will examine aspects of pediatric imageacquisition and segmentation along with other key methodology concerns. The purpose of thisreview is to demonstrate the novel applications of advanced MRI adiposity phenotypingapproaches in children.

OverviewMRI can be used to measure total body and regional tissue and organ volumes. The mass ofeach tissue or organ is then calculated as the product of volume and corresponding density (i.e.adipose tissue 0.92 kg/l, skeletal muscle 1.04 kg/l) [18]. Adipose tissue can be further dividedinto subcutaneous, visceral, and intramuscular adipose tissue [19]. Organ volumes that areusually quantified by MRI include liver, kidney, spleen, and brain [20]. Cardiac MRI can beused to separately study heart mass and function. The mass of these metabolically active organsaccurately predicts resting energy expenditure in adults [20], but the energy cost of growthrequires additional model development for estimation of resting energy expenditure in children[21].

Usually whole body magnetic resonance T1-weighted images are acquired as axial images with5–10 mm thickness and 40 mm or less between-slice intervals. Due to both the long analysistime and the incurred high cost, contiguous whole body MRI is rarely recommended for use,except in infants [22,23], who require much fewer slices. The subject is usually positionedeither prone or supine with their arms stretched above their head. Due to the limited magnetbore length, the upper and lower body are scanned separately in adults with repositioning usingthe L4–L5 intervertebral disc as the point of origin. In young children this repositioning maynot be necessary. Some technician training will facilitate acquisition of the whole bodyprotocol, which is different from most clinical diagnostic scans. It is also very important tokeep the body within the field of view, especially when scanning obese subjects.

Once acquired, MRI and CT scans must be analyzed by the process referred to as imagesegmentation. There are various segmentation programs available, both commercial anddeveloped in-house. The segmentation procedure for MRI is partially automated, althoughexpert analyst input is still required.

Generally, the segmentation tools can be divided into three groups: manual delineation; low-level segmentation such as thresholding and region growing; and model-based segmentationmethods [24]. Specific technical considerations are required for analyzing each tissue andorgan. For example, analyzing small adipose tissue depots requires consideration of theregional threshold, particularly when inhomogeneity of the magnetic field is present. On theother hand, organ analysis is carried out by manual tracing in most cases and knowledge ofanatomy is required.

Shen et al. Page 2

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Segmentation of a ‘whole-body’ into subcutaneous, visceral, and intramuscular adipose tissue,skeletal muscle, and ‘residual’ usually requires about 4–8 h with a slice gap of 5 cm (i.e. 40slices) and three to five working days for contiguous MRI scans (i.e. 200 slices), dependingon the number of acquired slices and the expertise of the analyst. MRI segmentation solely foradipose tissue requires less time than for segmenting a scan for all compartments. The analyst’straining includes use of segmentation software and knowledge of cross-sectional anatomy.Moreover, it is important for the analyst to recognize some commonly observed artifacts thatcould influence the differentiation of tissues. These include motion artifacts caused bybreathing or bowel peristaltic movements, blood flow artifacts, and other sources of imagedistortion. Some scanners also have magnetic field inhomogeneity. Although artifacts can bereduced or eliminated by setting specific scan parameters and breath holding, analysts shouldbe trained to recognize the presence of image artifacts.

Most image analysis software now offers three-dimensional reconstruction based on imagesegmentation data (Fig. 1). Three-dimensional reconstructions are very useful fordemonstration purposes. Investigators can visually sense not only the size and distribution oftissue compartments, but also longitudinal changes in tissue/organ size and distribution.

Quality controlLike all measurement procedures, quality control is an important aspect of image analysis.Reproducibility, such as intra-analyst variation, should be routinely evaluated. As large studiesmay require analysis by more than one technician over a relatively long time, it is essential toevaluate inter-analyst variation and to explore for any time drift. In adults, the intra-classcorrelation coefficient among trained analysts is reported – skeletal muscle 0.99 (0.89, 1.0);subcutaneous adipose tissue 0.99 (0.81, 1.0); visceral adipose tissue 0.95 (0.58, 0.99) [25] –and these values indicate a high degree of absolute agreement across image readers. Thevariation in repeated measurements is larger for intramuscular adipose tissue [26,27] and useof high resolution MRI in the future may improve segmentation of this compartment [28].Adipose tissue within small anatomic areas has a large measurement error [29] and tissuecompartments are much smaller in children than in adults. Reproducibility data should beacquired in future pediatric studies to provide more reliable power calculations and assessmentof pediatric data accuracy.

As noted earlier, whole body MRI scans are always acquired by slices with a gap of variablemagnitude. The volume between two adjacent slices is then estimated by geometric models[30]. There are differences between the measured and the true volumes but the difference isusually within 5% when the slice gap is below 40 mm [30]. Errors caused by slice gap insubcutaneous adipose tissue and skeletal muscle are similar in children and adults [31,32].Slice gaps cause larger errors in visceral adipose tissue volume estimates in children than inadults, but these errors can be compensated by studying large numbers of subjects [32]. WhileChen et al. [32] included children of ages 5–17 years in their studies, future studies need toexamine measurement errors in groups with a narrower age range.

In addition to volume quantification of tissue compartments by whole-body imaging protocols,tissue compartments, especially visceral and abdominal subcutaneous adipose tissue, are oftenreported as the area of a single slice. Because of the cost of whole body scans and concernsover radiation exposure with CT studies, single-slice MRI scans have often been used eventhough they are less accurate than whole body scans for component estimation [33●,34●]. It isimportant to recognize that single-slice studies only provide an area when reporting tissuecompartments, in contrast to the volumes reported in multiple slice and whole body MRIstudies. The single-slice CT and MRI studies are usually performed at the L4–L5 level. In both

Shen et al. Page 3

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

adults [33●] and children [35], however, a slice in the abdomen higher than the L4–L5 levelhas been shown to represent visceral adipose tissue better than that at the L4–L5 level.

Magnetic resonance imaging of infantsThe nutritional status during infancy and in utero have been linked to obesity and health risksin later life [36,37]. MRI is a well-suited method for evaluating infants as no radiation exposureis present. As infants are usually below 70 cm in length, they are scanned by contiguous MRIscan protocols instead of discrete slices and the required scan time is usually less than 10 min[22,23]. On the other hand, as the tissue structures in infants are small, higher resolution isrequired than in adults for adequate accuracy. This is usually achieved by using receive coilsin addition to the built-in scanner body coil used in adult studies.

Although there are no known risks of MRI, monitoring of infants and some protectiveprocedures are essential. To avoid artifacts caused by motion, infants are ideally scannedsleeping after being fed. A pulse oximeter is often attached to the fingers or toes of infants andthe MRI technician can then monitor blood oxygen saturation from the workstation. Afluctuation in oxygen saturation levels occurs when the infant cries. Infants should also beprotected from the scanner noise by using infant-specific earplugs. As most MRI scanner roomstend to have a low temperature because of the liquid helium or liquid nitrogen in thesuperconducting magnet, it is important to keep the infant warm with blankets. In fact, untilrecently, premature infants could not be scanned due to the low room temperature of the MRIsuite. Recent development of the MRI compatible incubator, however, opens the potential forstudying the nutritional status of premature infants [38].

Magnetic resonance spectroscopyIntramyocellular lipid (IMCL) and intrahepatic lipid (IHL) have been increasingly measuredin children by non-invasive proton (or 1-hydrogen) magnetic resonance spectroscopy (MRS)[39]. Both lipid compartments are closely related to insulin resistance [40–42] and thesensitivity and repeatability of these approaches were established in recent patient studies[43–45]. The utility and practicality of MRS are further advanced by the availability of threetesla clinical magnetic resonance scanners and other whole-body high-field magneticresonance scanners in hospitals and research centers. These spectroscopic methods requiremore advanced pre-scan adjustments, such as selection of anatomic location, shimming of localmagnetic field, and suppression of water signals. The quality of the obtained MRS data dependson the rigor of these pre-scan adjustments, which require investigator expertise and time. TheIMCL or IHL is quantified from corresponding spectral peaks through a numerical fitting/integration procedure and results are typically expressed as a ratio to creatine or water content.Currently, there are two main types of MRS approaches, single voxel spectroscopy (SVS) andchemical shift imaging (CSI).

Single voxel spectroscopyA single magnetic resonance spectrum is obtained with SVS from a localized volume of interestin liver parenchyma or muscle tissue. Two variations of the technique are generallyimplemented on magnetic resonance scanners. One method is ‘point resolvedspectroscopy’ (PRESS), and the other is ‘stimulated echo acquisition mode’ (STEAM). Anumber of metabolites in the resulting muscle spectrum, such as creatine, extramyocelluar lipid(EMCL) from infiltrated adipose tissue, and IMCL, exhibit well-defined spectral peaks at 3.0,1.5, and 1.3 ppm, respectively [44]. As noted, the spectral separation of EMCL and IMCLpeaks is small and the reliability of IMCL quantification depends on the sharpness of spectralline width, which reflects the shimming quality achieved prior to acquisition. Generally, toavoid contamination of lipid signals from infiltrated adipose tissue in muscle, especially in

Shen et al. Page 4

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

extremely obese and diabetic subjects, it is highly desirable to capture data from a small volumeof muscle (i.e. 1–3 cm3) [40,46]. The small volume leads to a reduced MRS signal and a largenumber of signal averages are therefore typically necessary to reach a reasonable signal tonoise ratio. Fortunately, children usually do not have much adipose tissue infiltration in theirmuscles. On the other hand, muscle size in children is much smaller than that of adults andsome MRI scanners may not be able to scan a small voxel within a single muscle of youngchildren.

Signals of hepatic lipid in children can be detected and quantified similarly as in adults. Thethin abdominal wall in children also favors the use of a surface coil in the exam, which can beplaced in proximity to the liver. Because the surface coil significantly improves the signal tonoise ratio (SNR) of the resulting magnetic resonance spectrum, a single scan acquisitionwithout data averaging is often sufficient. If multiple acquisitions are needed to improve SNR,respiratory gating of data acquisition is recommended.

Chemical shift imagingCSI, a spectroscopic imaging method that may be available on some MRI scanners, can beperformed in one, two, or three spatial dimensions, and collects spectroscopic data from arelatively large volume in a spatially resolved manner. The small-yielded nominal size of multi-voxel MRS (i.e. 0.125 cm3) potentially solves the problem of scanning the small muscles inyoung children. With its large volume of coverage and high spatial resolution compared toSVS, CSI provides more information such as the lipid content of different muscles [47].Through a given spatial encoding scheme, CSI yields a set of spectra that are spatially resolvedover the intended volume. To resolve spectra spatially in practice, encoding magneticresonance signals in multiple dimensions is required using gradient pulses with linearincrements in a large number of steps, such as 16, 24, and so forth. This is why a CSI scan istypically slower than SVS methods (i.e. 15–20 min versus 3–6 min). CSI, however, providesmultiple, spatially-resolved spectra that often justifies the effort in practice. Also, CSI willlikely become faster in the near future with newly developed acquisition techniques thatcombine the CSI approach with the volume selection methods used with SVS.

Dixon methodsThe Dixon method, which was initially developed and demonstrated more than a decade ago[48], can provide quantitative information on regional tissue fat content. This method hasgreater potential use in research than pure image ‘segmentation’ methods that quantify onlytissue volumes. The fat content of a region of interest within an organ or of the entire organcan be obtained with processing of Dixon image data sets. The Dixon method is not only usefulfor separating muscle from fat tissues, but is also applicable for evaluating liver and evenpancreas fat content [49●]. Although not quantifying fat solely within β cells, the quantificationof pancreas fat by the Dixon method is unique, as spectroscopy methods usually cannot beapplied for the pancreas that is located deep in the abdomen. The Dixon approach generallycollects anatomical images with water and fat signals encoded differently in two or threeimaging acquisitions [50]. Exploiting chemical shift differences between water and fatmagnetic resonance signals, a Dixon method selects several gradient echo times (TEs) to allowthe evolution of magnetic resonance signals according to their chemical shifts. Water and fat-only images are calculated from these resulting Dixon images. The drawback of the Dixonmethod is that it requires longer scanning times, post processing, and the method is not currentlyavailable on all clinical MRI scanners.

Shen et al. Page 5

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Cardiac magnetic resonance imagingCardiac MRI is now being used along with adipose tissue measurements in pediatrics, ascardiovascular disease is a major consequence of obesity. Cine MRI is a clinically availabletechnique that can assess the physiology and function of the heart. Accurate assessment ofventricular mass, stoke volume, and ejection fraction is a major strength of cine MRI. Taggingof cardiac tissue and blood, a unique feature of MRI compared with other cardiac measurementtechnologies, enables the evaluation of regional wall motion and velocity mapping of bloodflow [51]. Early detection of changes in cardiovascular function, such as aortic elasticity, is ofresearch interest in children. In young children that cannot hold their breath, cardiacsynchronization and respiratory gating are both necessary [52]. State-of-the-art MRI equipmentnow permits many magnetic resonance acquisitions to be completed in just a few seconds[52] and some ultra-fast imaging methods in combination with sensitivity encoding (SENSE)are necessary [53].

Potential pediatric magnetic resonance imaging problems and solutionsCompared with adults, it is usually difficult for children to stay immobile within the MRIsystem and to follow instructions. The problem of restlessness can at least be partially solvedby having children play beforehand in a simulated MRI scanner. Video games for educatingchildren in following specific instructions have also been developed and proven successful inteaching them the study process and how to cooperate. Nevertheless, there remains difficultyin scanning children between the ages of 6 months and 4 years. Most children in this age groupare unable to remain still and to understand instructions. Sedation, for ethical reasons, is usuallynot used on children participating for research purposes.

Recent advances in MRI technology provide a potential solution to the problem of restlessnessand motion during the scan. The Periodically Rotated Overlapping Parallel Lines withEnhanced Reconstruction (PROPELLER) data acquisition has proven successful in reducingartifacts caused by movement in only one direction at a time [54]. This technique reducesmovement artifacts in pediatric brain scans [55], although there remains a need to developtechniques that can compensate artifacts induced by muscle movement or breathing that causeorgans and soft tissues to move simultaneously in different directions. Recently, 1.5 T MRIscanners with bore diameters up of to 70 cm have been developed [56]. Wide bore scannersprovide the option for the child’s mother to lie with the subject during the scan.

A special coil design might be needed for adequate resolution of scans in children. While headcoils have been used in scanning the whole body of infants, these coils are not long enough tocover the whole body of most children. To our knowledge there is no single coil designed toscan the whole body of toddlers. Although pediatric coils have been built for scanning the spinetogether with the head of children, these coils are not long enough to cover the whole body andcannot generate images with adequate signals of the anterior side of the body.

Protocol selectionThere are several factors that influence the selection of an MRI protocol for a specific study.Cost and benefit is a key consideration as image segmentation or postprocessing is expensivefor most investigators outside of specialized center. For example, scanning a subject with arare, weight-related genetic disease should include both total body adipose tissue and adiposetissue distribution. This will require 40 to 50 separate images in a typical adult scanned witha 4 cm slice gap. Similarly, evaluating the effects of a new weight control drug may requiredetection of small total body and regional adipose tissue changes that are best captured inlongitudinal whole body scans of a small number of experimental participants. On the otherhand, for large-scale studies with adequate power and restricted budgets or for children that

Shen et al. Page 6

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

cannot remain immobile long enough inside the MRI scanner, single slice MRI or a few MRIslices are the appropriate protocol choice.

Another consideration is subject age and maturity. For infants and young children, whole bodycontiguous MRI scans should be considered as the total body length is relatively small. Thescan can therefore be completed quickly and the analysis time is reasonable due to the smalloverall slice number. On the other hand, well-developed adolescents can be scanned byprotocols similar to that of adults. A smaller slice gap than used in adults can be applied in pre-pubertal children without having much impact on the total body compartment estimates.

Protocol consideration is important when scanning growing children over time in longitudinalstudies. Adipose tissue distribution can change dramatically, especially during puberty, andthus multi-slice whole body MRI scans would avoid missing fat loss at one site and gain inanother. Another problem in children is landmark identification with elongation of bonesduring growth. In comparison, anatomic sites remain relatively stable in adults over long timeperiods. For example, when evaluating calf IMCL in adults, the imaging location is markedby the site distance from the knee. This distance changes in children as they grow. Thus,measurement from multiple bony landmarks and calculation of the relative distance willprovide a more reproducible means of site identification in longitudinal pediatric studies.Measuring variation within a muscle will also help to evaluate the effects of positioning onIMCL variation. If IMCL variation within a muscle is negligible, the exact anatomical locationis therefore likely unimportant.

ConclusionMRI scanners are widely available and the acquired data in pediatric studies safely provideunique nutritional and metabolic information not obtainable by other methods. Amisconception is that scanning times are long, complex and require substantial interruption ofusual clinical patient flow. Once acquired, however, the present post-processing andsegmentation procedures are more limited. Educational and consulting resources are availableat specialized centers, thus facilitating MRI and MRS as powerful pediatric research tools.Moreover, advances in MRI technology, including rapid imaging sequences with morechannels for data acquisition along with advanced coil designs and motion artifact corrections,will improve magnetic resonance capabilities for use in infants and young children.

ReferencesPapers of particular interest, published within the annual period of review have beenhighlighted as

● of special interest

●● of outstanding interest

Additional references related to this topic can also be found in the Current World Literaturesection in this issue (p. 651).

1. Gortmaker SL, Dietz WH Jr, Sobol AM, Wehler CA. Increasing pediatric obesity in the United States.Am J Dis Child 1987;141:535–540. [PubMed: 3578167]

2. Troiano RP, Flegal KM, Kuczmarski RJ, et al. Overweight prevalence and trends for children andadolescents. The National Health and Nutrition Examination Surveys, 1963 to 1991. Arch PediatrAdolesc Med 1995;149:1085–1091. [PubMed: 7550810]

3. Rocchini AP. Childhood obesity and a diabetes epidemic. N Engl J Med 2002;346:854–855. [PubMed:11893799]

Shen et al. Page 7

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

4. Strauss RS, Pollack HA. Epidemic increase in childhood overweight, 1986–1998. JAMA2001;286:2845–2848. [PubMed: 11735760]

5. Williams DP, Going SB, Lohman TG, et al. Body fatness and risk for elevated blood pressure, totalcholesterol, and serum lipoprotein ratios in children and adolescents. Am J Public Health 1992;82:358–363. [PubMed: 1536350]

6. Weiss R, Dziura J, Burgert TS, et al. Obesity and the metabolic syndrome in children and adolescents.N Engl J Med 2004;350:2362–2374. [PubMed: 15175438]

7. Rosenbaum M, Nonas C, Horlick M, et al. beta-Cell function and insulin sensitivity in earlyadolescence: association with body fatness and family history of type 2 diabetes mellitus. J ClinEndocrinol Metab 2004;89:5469–5476. [PubMed: 15531499]

8. Serdula MK, Ivery D, Coates RJ, et al. Do obese children become obese adults? A review of theliterature Prev Med 1993;22:167–177.

9. Guo SS, Roche AF, Chumlea WC, et al. The predictive value of childhood body mass index valuesfor overweight at age 35 y. Am J Clin Nutr 1994;59:810–819. [PubMed: 8147324]

10. Nieto FJ, Szklo M, Comstock GW. Childhood weight and growth rate as predictors of adult mortality.Am J Epidemiol 1992;136:201–213. [PubMed: 1415142]

11. Must A, Jacques PF, Dallal GE, et al. Long-term morbidity and mortality of overweight adolescents.A follow-up of the Harvard Growth Study of 1922 to 1935. N Engl J Med 1992;327:1350–1355.[PubMed: 1406836]

12. Goran MI. Measurement issues related to studies of childhood obesity: assessment of bodycomposition, body fat distribution, physical activity, and food intake. Pediatrics 1998;101:505–518.[PubMed: 12224657]

13. Weiss R, Dufour S, Taksali SE, et al. Prediabetes in obese youth: a syndrome of impaired glucosetolerance, severe insulin resistance, and altered myocellular and abdominal fat partitioning. Lancet2003;362:951–957. [PubMed: 14511928]

14. Brambilla P, Manzoni P, Sironi S, et al. Peripheral and abdominal adiposity in childhood obesity. IntJ Obes Relat Metab Disord 1994;18:795–800. [PubMed: 7894517]

15. Brambilla P, Manzoni P, Agostini G, et al. Persisting obesity starting before puberty is associatedwith stable intraabdominal fat during adolescence. Int J Obes Relat Metab Disord 1999;23:299–303.[PubMed: 10193876]

16. Simha V, Garg A. Phenotypic heterogeneity in body fat distribution in patients with congenitalgeneralized lipodystrophy caused by mutations in the AGPAT2 or seipin genes. J Clin EndocrinolMetab 2003;88:5433–5437. [PubMed: 14602785]

17. Pichiecchio A, Uggetti C, Egitto MG, et al. Quantitative MR evaluation of body composition inpatients with Duchenne muscular dystrophy. Eur Radiol 2002;12:2704–2709. [PubMed: 12386760]

18. Snyder, WM.; Cook, MJ.; Nasset, ES., et al. Report of the Task Group on Reference Man, ICRPpublication 23. Oxford: Pergamon Press; 1975.

19. Shen W, Wang ZM, Punyanita M, et al. Adipose tissue quantification by imaging methods: a proposedclassification. Obes Res 2003;11:5–16. [PubMed: 12529479]

20. Gallagher D, Belmonte D, Deurenberg P, et al. Organ-tissue mass measurement allows modeling ofREE and metabolically active tissue mass. Am J Physiol Endocrinol Metab 1998;275:E249–E258.[PubMed: 9688626]

21. Hsu A, Heshka S, Janumala I, et al. Larger mass of high-metabolic-rate organs does not explain higherresting energy expenditure in children. Am J Clin Nutr 2003;77:1506–1511. [PubMed: 12791631]

22. Harrington TA, Thomas EL, Modi N, et al. Fast and reproducible method for the direct quantitationof adipose tissue in newborn infants. Lipids 2002;37:95–100. [PubMed: 11878317]

23. Olhager E, Thuomas KA, Wigstrom L, Forsum E. Description and evaluation of a method based onmagnetic resonance imaging to estimate adipose tissue volume and total body fat in infants. PediatrRes 1998;44:572–577. [PubMed: 9773848]

24. Rogowska J, Batchelder K, Gazelle GS, et al. Evaluation of selected two-dimensional segmentationtechniques for computed tomography quantitation of lymph nodes. Invest Radiol 1996;31:138–145.[PubMed: 8675421]

25. Heshka, S.; Punyanitya, M.; Shen, W., et al. Inter-reader reliability in reconstructing organ/tissuevolumes from magnetic resonance images [abstract]. In: Marchesi, VT., editor. Experimental

Shen et al. Page 8

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Biology. Washington, DC. Washington DC: Federation of American Societies for ExperimentalBiology; 2004 Apr. 2004 2004 p. 17-21.

26. Heshka, S.; Gallagher, D.; Heymsfield, S., et al. Precision of MRI body composition measurements[abstract]. Seventh International Symposium In Vivo Body Composition Studies: Linking Structureand Function. (BC2005); 7–9 September 2005; Southampton. Southampton: University ofSouthampton; 2005.

27. Janssen I, Fortier A, Hudson R, Ross R. Effects of an energy-restrictive diet with or without exerciseon abdominal fat, intermuscular fat, and metabolic risk factors in obese women. Diabetes Care2002;25:431–438. [PubMed: 11874926]

28. Shen, W.; Punyanita, M.; Wu, K., et al. High resolution magnetic resonance imaging in quantificationof intra- and peri-muscular adipose tissue in the leg region [abstract]. The North AmericanAssociation for the Study of Obesity 2003 annual scientific meeting; 11–15 October 2003; FortLauderdale. Florida. Silver Spring: North American Association for the Study of Obesity;

29. Elbers JM, Haumann G, Asscheman H, et al. Reproducibility of fat area measurements in young,non-obese subjects by computerized analysis of magnetic resonance images. Int J Obes Relat MetabDisord 1997;21:1121–1129. [PubMed: 9426379]

30. Shen W, Wang Z, Tang H, et al. Volume estimates derived in vivo by imaging methods: modelcomparisons with visible woman as the reference. Obes Res 2003;11:217–225. [PubMed: 12582217]

31. Asare, NK.; Shen, W.; Punyanitya, M., et al. Tissue volume estimates derived in vivo by magneticresonance imaging (MRI): validity of 5 cm slice protocol in children. The North AmericanAssociation for the Study of Obesity 2004 annual scientific meeting; 14–18 November 2004; LasVegas. Silver Spring: North American Association for the Study of Obesity; 2004.

32. Chen, J.; Shen, W.; Punyanitya, M., et al. Tissue volume estimates derived by magnetic resonanceimaging (MRI): validity of different slice interval protocols in children [abstract]. The NorthAmerican Association for the Study of Obesity 2005 annual scientific meeting; 15–19 October 2005;Vancouver. Silver Spring: North American Association for the Study of Obesity;

33●. Shen W, Punyanitya M, Wang Z, et al. Visceral adipose tissue: relations between single-slice areasand total volume. Am J Clin Nutr 2004;80:271–278. [PubMed: 15277145]This article identifiedthat an image slice acquired at upper abdomen is better than the traditionally used L4–L5 slice inrepresenting total visceral adipose tissue amount

34●. Shen W, Punyanitya M, Wang Z, et al. Total body skeletal muscle and adipose tissue volumes:estimation from a single abdominal cross-sectional image. J Appl Physiol 2004;97:2333–2338.[PubMed: 15310748]This article is the first to report the best abdominal single slice in representingtotal adipose tissue and muscle volume in a relatively large healthy subject pool

35. Shen, W.; Punyanitya, M.; Chen, J., et al. Visceral adipose tissue: relationships between single sliceareas and total volume in children [abstract]. Seventh International Symposium In Vivo BodyComposition Studies: Linking Structure and Function. BC2005; 7–9 September 2005; Southampton.Southampton: University of Southampton; 2005. p. O-23

36. Schaefer-Graf UM, Pawliczak J, Passow D, et al. Birth weight and parental BMI predict overweightin children from mothers with gestational diabetes. Diabetes Care 2005;28:1745–1750. [PubMed:15983329]

37. Boney CM, Verma A, Tucker R, Vohr BR. Metabolic syndrome in childhood: association with birthweight, maternal obesity, and gestational diabetes mellitus. Pediatrics 2005;115:e290–e296.[PubMed: 15741354]

38. Veeraraghavan, S.; Miller, SP.; Mukherjee, P., et al. Improved diffusion MR imaging of prematurenewborns using an MR compatible incubator and a specialized neonate MR coil [abstract].International Society for Magnetic Resonance in Medicine 13th scientific meeting; 7–13 May 2005;Miami. Berkeley: International Society for Magnetic Resonance in Medicine; 2005. p. 1253

39. Boesch C, Slotboom J, Hoppeler H, Kreis R. In vivo determination of intra-myocellular lipids inhuman muscle by means of localized 1H-MR-spectroscopy. Magn Reson Med 1997;37:484–493.[PubMed: 9094069]

40. Perseghin G, Scifo P, Cobelli FD, et al. Intramyocellular triglyceride content is a determinant of invivo insulin resistance in humans: a 1H-13C nuclear magnetic resonance spectroscopy assessmentin offspring of type 2 diabetic parents. Diabetes 1999;48:1600–1606. [PubMed: 10426379]

Shen et al. Page 9

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

41. Sinha R, Dufour S, Petersen KF, et al. Assessment of skeletal muscle triglyceride content by (1)Hnuclear magnetic resonance spectroscopy in lean and obese adolescents: relationships to insulinsensitivity, total body fat, and central adiposity. Diabetes 2002;51:1022–1027. [PubMed: 11916921]

42. Tiikkainen M, Tamminen M, Hakkinen AM, et al. Liver-fat accumulation and insulin resistance inobese women with previous gestational diabetes. Obes Res 2002;10:859–867. [PubMed: 12226133]

43. Szczepaniak LS, Nurenberg P, Leonard D, et al. Magnetic resonance spectroscopy to measure hepatictriglyceride content: prevalence of hepatic steatosis in the general population. Am J PhysiolEndocrinol Metab 2005;288:E462–E468. [PubMed: 15339742]

44. Szczepaniak LS, Babcock EE, Schick F, et al. Measurement of intracellular triglyceride stores by Hspectroscopy: validation in vivo. Am J Physiol Endocrinol Metab 1999;276:E977–E989.

45. Shen, W.; Mao, X.; Wolper, C., et al. Reproducibility of measurements of intramyocellular lipid levelsby single- and multi-voxel proton MRS in overweight and lean subjects during controlled dietary fatintake [abstract]. International Society for Magnetic Resonance in Medicine 13th scientific meeting;7–13 May 2005; Miami Beach. Berkeley: International Society for Magnetic Resonance in Medicine;2005.

46. Boden G, Lebed B, Schatz M, et al. Effects of acute changes of plasma free fatty acids onintramyocellular fat content and insulin resistance in healthy subjects. Diabetes 2001;50:1612–1617.[PubMed: 11423483]

47. Vermathen P, Kreis R, Boesch C. Distribution of intramyocellular lipids in human calf muscles asdetermined by MR spectroscopic imaging. Magn Reson Med 2004;51:253–262. [PubMed:14755649]

48. Dixon WT. Simple proton spectroscopic imaging. Radiology 1984;153:189–194. [PubMed: 6089263]49●. Kovanlikaya A, Mittelman SD, Ward A, et al. Obesity and fat quantification in lean tissues using

three-point Dixon MR imaging. Pediatr Radiol 2005;35:601–607. [PubMed: 15785930]This studyused the Dixon method to quantify ectopic fat in children and is novel in quantifying pancreas fat

50. Glover GH. Multipoint Dixon technique for water and fat proton and susceptibility imaging. J MagnReson Imaging 1991;1:521–530. [PubMed: 1790376]

51. Fogel MA. Assessment of cardiac function by magnetic resonance imaging. Pediatr Cardiol2000;21:59–69. [PubMed: 10672615]

52. Vick GWR. Recent advances in pediatric cardiovascular MRI. Curr Opin Pediatr 2003;15:454–462.[PubMed: 14508292]

53. Sodani G, Sciuto F, Di Roma M, et al. Cardiac MRI with SENSE technology: study of ventricularand valvular functionality in 30 patients. Radiol Med (Torino) 2004;107:1–10. [PubMed: 15031692]

54. Forbes KP, Pipe JG, Bird CR, Heiserman JE. PROPELLER MRI: clinical testing of a novel techniquefor quantification and compensation of head motion. J Magn Reson Imaging 2001;14:215–222.[PubMed: 11536397]

55. Forbes KP, Pipe JG, Karis JP, et al. Brain imaging in the unsedated pediatric patient: comparison ofperiodically rotated overlapping parallel lines with enhanced reconstruction and single-shot fast spin-echo sequences. AJNR Am J Neuroradiol 2003;24:794–798. [PubMed: 12748073]

56. Simens Medical Solutions. MAGNETOM Espree: the first open bore MRI with Tim. InternationalSociety for Magnetic Resonance in Medicine 13th scientific meeting; 7–13 May 2005; Miami.Berkeley: International Society for Magnetic Resonance in Medicine; 2005. p. ix

AbbreviationsCSI

chemical shift imaging

CT computed tomography

IMCL intramyocellular lipid

MRI

Shen et al. Page 10

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

magnetic resonance imaging

SVS single voxel spectroscopy

Shen et al. Page 11

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1.Magnetic resonance scan and three-dimensional reconstruction of the different tissuecompartments of a healthy 16-year-old girl

Shen et al. Page 12

Curr Opin Clin Nutr Metab Care. Author manuscript; available in PMC 2007 June 19.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript