123.full.pdf

8/14/2019 123.full.pdf

1/8

Whole-body adipose tissue analysis: comparison of MRI, CT and

dual energy X-ray absorptiometry

1J KULLBERG, MS, 2J BRANDBERG, MD, 3J-E ANGELHED, PhD, 1H FRIMMEL, PhD, 3E BERGELIN, RN,

3L STRID, RN, 1H AHLSTROM, MD, PhD, 1L JOHANSSON, PhDand 2,3,4L LONN, MD, PhD

1Department of Radiology, Uppsala University Hospital, Uppsala, 2Department of Radiology, Sahlgrenska University

Hospital, Goteborg, 3Department of Metabolism and Cardiovascular Research, The Sahlgrenska Academy at Goteborg

University, Goteborg, Sweden and 4Faculty of Health Sciences, Radiology and Vascular Surgery, Copenhagen University,

Denmark

ABSTRACT.The aim of this study was to validate a recently proposed MRI-based T1-mapping method for analysis of whole-body adipose tissue (AT) using an established CTprotocol as reference and to include results from dual energy X-ray absorptiometry(DEXA). 10 subjects, drawn from the Swedish Obese Subjects Sibling-pairs study, wereexamined using CT, MRI and DEXA. The CT analysis was based on 28 imaged slices. T1

maps were calculated using contiguous MRI data from two different gradient echosequences acquired using different flip angles. CT and MRI comparison was performedslice-wise and for the whole-body region. Fat weights were compared between allthree modalities. Strong correlations (r>0.977,p,0.0001) were found between MRIand CT whole-body and AT volumes. MRI visceral AT volume was underestimated by0.790.75 l (p50.005), but total AT was not significantly different from that estimatedby CT (MRI CT 5 0.611.17 l; p50.114). DEXA underestimated fat weights by5.231.71 kg (p50.005) compared with CT. MRI underestimated whole-body volumeby 2.031.61 l (p50.005) compared with CT. Weights estimated either by CT or byDEXA were not significantly different from weights measured using scales. Inconclusion, strong correlations were found between whole-body AT results from CT,MRI-basedT1 mapping and DEXA. If the differences between the results from T1-mapping and CT-based analysis are accepted, the T1-mapping method allows fullyautomated post-processing of whole-body MRI data, allowing longitudinal whole-body studies that are also applicable for children and adolescents.

Received 26 November2007Revised 29 February 2008Accepted 11 April 2008

DOI: 10.1259/bjr/80083156

2009 The British Institute of

Radiology

The prevalence of overweight and obese individuals israpidly increasing in many countries around the world.Both the amount and the distribution of adipose tissue(AT) are associated with premature death [1]. Therefore,there is an increasing need for accurate and automatedtools in the field of body composition.

CT is an imaging modality that allows rapid acquisi-tion and accurate body composition analysis [2, 3].However, in order to reduce the exposure to ionizingradiation, only a limited number of image slices can be

used. A well-established and validated CT-based whole-body analysis method that uses 28 slices to divide thebody into 12 main compartments of tissues, organs andgas has previously been described [2]. Body weightsestimated using this method show only minor deviationsfrom real body weights.

Dual energy X-ray absorptiometry (DEXA) was ori-ginally constructed for bone density analysis. However,it can also be used in studies of body composition [4, 5].DEXA body composition analysis gives the total weightsof fat, lean tissue and bone mineral content as output. Atotal body DEXA investigation exposes the subject to

minimal amounts of radiation (#15 mSv) [6]. As thescanner does not have a gantry, the patient compliancelevel is high, with respect to claustrophobia.

MRI allows the analysis of body composition [3, 79]without any known long-term side effects, allowing largecoverage, repeated acquisition and studies of childrenand adolescents [10]. Whole-body MR analysis ismotivated by the need for accurate phenotype determi-nations. Age and ethnicity have an impact on bodycomposition; hence, studies of these factors probably

gain from extensive analysis. Larger coverage, moredense sampling and repeated acquisitions demandautomation of the data processing. Automation iscomplicated because MRI intensity levels are given inarbitrary units (AUs) and images are often affected byintensity inhomogeneities. An MRI-based T1-mappingtechnique, hereafter denoted as the MRI method, hasrecently been shown to simplify automated analysis ofAT from whole-body data [11]. However, the techniquehas not yet been validated in a contiguous whole-bodyregion nor have the results been compared with thosefrom previously validated modalities.

The objective of this study was to validate the whole-

body MRI method using the validated CT-based whole-body method as reference. Whole-body DEXA analysiswas also performed. MRI, CT and DEXA results were

Address correspondence to: Joel Kullberg, MRT, Entrance 24,Uppsala University Hospital, SE-751 85 Uppsala, Sweden. E-mail:[email protected]

The British Journal of Radiology, 82 (2009), 123130

The British Journal of Radiology, February 2009 123

8/14/2019 123.full.pdf

2/8

compared on a whole-body level, as well as on a slice-wise level (MRI and CT only).

Methods and materials

Subjects and anthropometric measurements

10 volunteer subjects (see Table 1) were scanned usingthe CT-based 28-slice protocol [2], the MRI protocol [11]and the DEXA protocol [12]. The subjects were membersof two nuclear families and also participated in theSwedish Obese Subjects (SOS) Sibling-pairs study. TheSOS Sibling-pairs study is an extension of the SOS study,which is a longitudinal study aiming to investigate thehealth effects from weight loss surgery [13, 14]. The focusof the Sibling-pairs study is to find relationships betweengenotypes and phenotypes.

This study was approved by the local ethics andradiation committees and all subjects gave informedconsent. We certify that all applicable institutional andgovernmental regulations concerning the ethical use ofhuman volunteers were followed during this research.

Height and weight were measured using standardizedequipment. The coefficient of variation values were

below 1% for these measurements. The acquisition ofthe CT, MRI and DEXA data from each subject wasperformed at the same site within a 3 h interval. Patientswere asked to fast for 5 h prior to scanning. Thepositioning of the subjects during the CT and MRIacquisitions was as similar as possible to minimizeintermodality positioning differences. The similarity ofposition was achieved by use of experienced personneland identical cushioning. The shapes of the CT and MRI

table-tops differed slightly.

CT hardware and reference protocol

The CT scanner (HiSpeed Advantage, version RP2;General Electric, Milwaukee, WI) was used to acquire 28axial images from subjects in the supine position(Figure 1). In addition to the 28 image slices, twopositions were determined from the scouts in order tomeasure the total length of the subjects in the supineposition. Positions were numbered 130 (including the28 scans: positions 229) from toes to finger tips(Figure 1). The exact positions of the scans were obtainedusing the scout images and have previously beendescribed [2]. Owing to limitations in the hardware, theslices were acquired using one repositioning. The slicesfrom the ankle joint to the upper border of the crista(positions 211) were first acquired. The subject was then

repositioned, after which the slices from the L3 level tothe wrists (positions 1229) were acquired. Obesesubjects that did not fit into the field of view (FOV)were repositioned to exceed only the FOV on one side ofthe body. Thus, the other side of the body, including atleast 50% of all tissue areas, was scanned and analysed.

Scan parameters were 120 kVp, 40240 mAs depend-

ing on body part and body size, 480 mm FOV, 5 mmslice thickness and 256 6 256 matrix size. The chargewas set according to a dose reduction scheme [15]. Theaverage effective dose per examination was 0.63 mSv(maximum 1.61 mSv).

CT analysis

The acquired CT data was semi-automatically ana-lysed using software developed at the Department ofMedicine, Goteborg University, Sweden. Measurementsof distance, circumference and area, and calculation ofvolumes, have previously been described in detail [2].

Calculation of the distance between slice 11 and slice 12was made by manual determination of the position of acommon skeletal feature in both half-body scouts. Theanalysis subdivides each scanned slice in subareas oftissues, organs and gas [2]. Subcutaneous adipose tissue(SAT) and visceral adipose tissue (VAT, i.e. the sum ofintra- and retro-peritoneal AT) were separated usingsemi-automated delineation. VAT was measured, whenpresent, between the symphysis upper border andlowest diaphragm dome, i.e. in slices 718. The totalvolume of each tissue was calculated using the measuredtissue slice areas and the distances between the slicesusing Equation 1:

V~

Xn{1

i~1 di

aizaiz1

2 1

Equation 1 determines the total tissue volume V foreach tissue type fromnslices (n.1) by using the distancedibetween sliceiandi+1 and the measured tissue areasaianda i+1from slice i and i+1, respectively.

When the ipsilateral arm of a subject exceeded theFOV (in slices 2428), the measurements were performedin the contralateral arm, and the results were doubled.When slices 623 were affected, the measurements oftotal and SAT areas from the unaffected side weredoubled. The centre line used was manually determined.

Total subject weight and fat weight were estimated bythe use of density values reported in the literature [2](AT, 0.923 kg dm3). Precision errors (standard errors)for SAT and VAT from repeated interpretation havepreviously been determined as 0.5% and 1.2%, respec-tively [2]. The reproducibility of a CT protocol thatacquired and analysed two of the slices (thigh and L4:slices 5 and 11) from the 28-slice protocol used in thisstudy has recently been investigated in 50 obese subjectsat the site performing the CT analysis used in this study(Brandberg et al, unpublished data). The standard errorsreported from repeated interpretation were 0.5% and1.1%, and from repeated acquisition were 2.2% and 6.0%,

for SAT and VAT, respectively. The time required for thepost-processing of the CT data from one subject, usingthe semiautomatic software, was approximately 3 h.

Table 1. Subject characteristics

Females (n56) Males (n54)

Mean Range Mean Range

Age (years) 51 3870 57 4571Height (m) 1.65 1.611.71 1.80 1.771.83Weight (kg) 76.4 59.3113 105 92.4124

BMI (kg m2) 27.9 22.938.7 32.5 28.837.1

BMI, body mass index.

J Kullberg, J Brandberg, J-E Angelhed et al

124 The British Journal of Radiology, February 2009

8/14/2019 123.full.pdf

3/8

MRI hardware and protocol

The contiguous whole-body MRI acquisition wasperformed on a 1.5 T clinical MRI scanner (GyroscanNT; Philips Medical Systems, Best, the Netherlands) usinga spoiled T1 weighted gradient echo sequence [11]. Themain scan parameters were: repetition time, 177 ms; echotime, 2.3 ms; FOV, 530 mm; and slice thickness, 8 mm.

Three whole-body volumes were acquired. The first wasacquired using a flip angle of 80 and was denoted asFlip80 [11]. The Flip80 data were used for automaticsegmentation of the bodies and lungs. It was also used inthe visual selection of the MR slices that best correspondedto the acquired CT slices. Scan parameters were turned offto ensure a constant MR signal scaling. Two more whole-

body volumes were acquired using flip angles of 80 and30 . These volumes were denoted as Flip80off andFlip30off, respectively, and were used in the calculationof theT1relaxation map [11]. Owing to limitations in thehardware, the whole-body volumes needed to be acquiredhalf-body-wise using one subject repositioning. Theacquisition has previously been described in detail [11].

MRI analysis

Registration of the half bodies was performed aspreviously described [11], with the exception that theregistration in the feethead direction was manuallyperformed by an experienced radiologist.

As T1-mapped values outside the bodies and in thelungs are of no interest, an automated image processingalgorithm was used to segment bodies and lungs fromthe Flip80 data (see Appendix A). The total subjectvolumes were estimated from MRI using the segmentedwhole-body volumes.

AT was segmented from the T1-mapped data using

thresholds automatically derived from the whole-bodyT1 histograms using a commonly used method. Thehistograms were analysed by fitting of two Gaussianfunctions: one was fitted to the histogram AT peak andthe other to the lean tissue peak. The optimization wasperformed using the summed least squares criterion andthe downhill simplex method of Nelder and Mead [16].The intersection between the two Gaussian functionswas chosen as the T1threshold value.

SAT and VAT were separated by an experiencedradiologist using manual classification. VAT was classi-fied using binary AT results from the thresholded T1maps by use of ImageJ software (National Institute of

Health, Bethesda, MD) [17]. The total SAT and VATvolumes were estimated using the results from themanually classified binary AT data.

Visual selection of the 28 slices within positions 130from the continuous whole-body data was performed byan experienced radiologist. The slice selections wereperformed based on the congruence of anatomicalstructures, using the acquired CT slices as a reference.The MR slice selections were performed using the Flip80data and the resulting slice positions were used in theslice-wise evaluation.

The simple thresholding of the T1-mapped data

includes bone marrow (BM) in the AT volumesmeasured. Accurate exclusion of BM is difficult toachieve in many MR slices acquired using this protocol.However, to assess the effect on the SAT areas measuredin the comparison with CT, the BM was manuallysegmented by an experienced operator in eight slices (29) from the binary AT volumes.

DEXA analysis

The DEXA scanner used was a LUNAR DPX-L (LunarCo., Madison WI) with software version 1.35 and anextended analysis program for total body analysis

(LUNAR Radiation, Madison, WI). Body fat, lean tissuemass, bone mineral content and body weight wereassessed. Quality assurance tests were conducted on adaily basis. The DEXA hardware used in this study had

been validated in a previous study performed by Lantzet al [12]. However, a different version of the software(v1.31) was used. Repeated examinations of 10 females(body mass index, 22.11.6 kg m2), using the oldversions of the software, showed body fat and leantissue mass differences of 1.7% and 0.7%, respectively.

Whole-body and slice-wise comparisons

Total body, AT, SAT and VAT volumes from the MRImethod were evaluated using the volumes from the CT-

based analysis as a reference. Body weights estimated byCT and DEXA were evaluated using the body weightsfrom a scale as a reference. Fat weights estimated by MRIand DEXA were compared using the values derived byCT as a reference.

Slice-wise total body, SAT and VAT areas from MRIwere evaluated using CT as a reference. To visualize thedifferences in areas estimated by CT and the MRI-basedanalysis, the contiguous slices were standardized for

body size. The 28 selected MRI slice positions from onesubject were used as a template and the contiguous slice

results from all subjects were linearly interpolated(nearest neighbour) between these 28 slice positions.The slice-wise comparisons did not include DEXA.

Figure 1. Illustration of the slicepositions.

Whole-body adipose tissue comparison of MRI, CT and DEXA


8/14/2019 123.full.pdf

4/8

Statistical analysis

Linear correlations were studied and are reported ascorrelation coefficients (r-values). Differences wereinvestigated using the Wilcoxon signed-rank test.Bivariate plots and linear regression analysis were usedto study dependencies. p-Values of ,0.05 were consid-ered significant.

Results

Whole-body comparisons

Correlations, differences and linear dependenciesbetween CT, MRI and DEXA measurements are given

in Table 2. Strong correlations (r>0.977, p,0.0001) wereseen for all estimated volumes. The MRI analysis wasseen to overestimate SAT volumes, whereas VAT andtotal body volumes were underestimated. Total subjectweights estimated from the CT and the DEXA analyseswere not seen to differ from the weights assessed byscales. The fat weights estimated from the MRI analysis

did not differ significantly from the fat weights estimatedby CT. DEXA was found to underestimate the total fatweights compared with both CT and MRI. Bivariate plotsof whole-body volume differences between MRI and CTare given in Figures 24 for total body, SAT and VAT,respectively. The total fat weights measured by CT, MRIand DEXA for the 10 subjects are given in Table 3.

Slice-wise comparisons

The results from the slice-wise comparisons of totalarea, SAT and VAT between CT and MRI are displayedin Figure 5.

Total body areasTotal slice areas were often underestimated by the MRI-

based method compared with CT (see Figure 5a,c). Theabsolute slice area was underestimated in 11 slicepositions and overestimated in 2 slice positions. Linearregression on all absolute area differences showed a

significant dependence on slice area (MRI CT 50.053CT +19.1;r50.584,p,0.0001). The slice positionedat the top of the skull (position 27) was found tooverestimate the area from the MRI-based methodcompared with CT by more than 30 cm2 in eight of thesubjects.

Subcutaneous adipose tissue areasSAT areas were often overestimated by MRI compared

with CT (see Figure 5a,d). Significant differences werefound in 19 slice positions. Slices from MRI positioned atthe ankle joint, knee, pelvis, thorax and wrists werefound to overestimate the SAT areas, whereas slices at

the L3 level and at the level of the lower orbital border(slices 12 and 25) underestimated the SAT areas whencompared with CT. Linear regression on all differencesshowed a dependence on the SAT area measured by CT(MRI CT 5 0.041CT + 16.7;r50.161,p,0.0001).

The slice-wise differences in SAT after manual exclu-sion of BM are shown in Figure 5c (dotted line).

Table 2. Whole-body correlations, differences and linear dependencies between CT, MRI and DEXA measurements (n510)

Volume comparisons (MRI CT)Volume Correlation Difference (l) Linear dependence (l)

Total 0.998 22.031.61,p50.005 20.045CT+1.920,p50.054

AT 0.995 2

0.611.17,p5

0.114 2

0.004CT0.754, p5

0.916SAT 0.977 2.772.41,p50.007 20.007CT+2.563,p50.933VAT 0.987 20.790.75,p50.005 20.209CT+0.310,p50.002

Subject weight comparisonsModalities Correlation Difference (kg)

CT scales 1.000 0.400.60,p50.114DEXA scales 1.000 20.220.42,p50.114

Fat weight comparisonsModalities Correlation Difference (kg)

MRI CT 0.995 20.561.08,p50.114DEXA CT 0.990 25.231.71,p50.005DEXA MRI 0.979 24.672.38,p50.005

Correlations are given as correlation coefficients (r-values). All correlations were significant (p,0.0001). Differences are given asmean standard deviation. AT, adipose tissue (SAT +VAT); SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue;DEXA, dual energy X-ray absorptiometry.

Figure 2. Whole-body total volume differences (MRI CT)plotted as a function of CT volume measured (the reference).The linear dependence was found to be: 0.045CT + 1.920;r250.39;p50.054.



8/14/2019 123.full.pdf

5/8

Significant differences were seen in four slice positions(compared with five before exclusion of BM). SAT wasoverestimated in slices at the symphysis lower borderand acetabulum upper border (slices 6 and 8) afterexclusion of BM, whereas the slices at the ankle joint andcalf (slices 2 and 3) were underestimated after theexclusion of BM. The absolute differences in SAT areasmeasured between MRI and CT were significantlyreduced in slices 2, 4 and 69 by the exclusion of BM.

Visceral adipose tissue areasVAT areas were underestimated by MRI compared with

CT (Figure 5a,e). Significant differences were seen in eightslice positions. Slices between the upper border of theacetabulum and the L1 level (positions 813) and at thepositions of the lowest diaphragm dome and 7 mm abovethis position (slices 16 and 17) were underestimated byMRI compared with CT. Linear regression on all differ-ences showed a dependence on the VAT area measured byCT (MRI CT 5 20.155CT 0.459;r50.575,p,0.0001).

Affected slices

Two MRI slices used in the slice comparison wereaffected by artefacts or visible geometrical alternations. The

body of one subject mildly exceeded the FOV in position22. The affected area was estimated to be 0.5% of the totalin-slice body area. Position 25 in another subject wasaffected by metal, resulting in an artefact that caused areduction in the estimated skull area of approximately 18%.In the CT analysis, 42 slices were affected. Subjectsexceeded the FOV in positions 626 and 1627. The arms(positions 2427) were, in part, out of the FOV in sixsubjects. The DEXA examinations did not manage to fullyinclude the arms into the FOV in the three heaviest subjects.

Discussion

We have shown that results from a previouslypresented MRI method [11] and from DEXA analysis

give measurements of total and AT volumes and weightsthat correlate highly with the results from the whole-

body CT protocol used as reference. We have also shownthat the MRI method gives measurements of total ATvolumes with few differences from the whole-body CTprotocol. To our knowledge, no whole-body compar-isons between CT and MRI, and CT and DEXA, havepreviously been presented.

Total slice areas were seen to be underestimated by theMRI-based analysis compared with CT. Larger areasshowed larger absolute and relative underestimations,indicating that geometrical distortion in MRI might be acontributing factor. The overestimations of the total sliceareas in the calf and at the top of the skull are probably

caused by systematic differences in the slice selection.The automated segmentation algorithm might also becausing the differences in the slice from the top of theskull.

The MRI method was frequently seen to overestimatethe SAT areas measured when compared with CT. Aprobable cause is that BM is included in the MRIanalysis. Slice area differences for SAT measured bythe two methods were smaller after manual exclusion ofBM from the analysis (see Figure 5d). Figure 5a,d alsoshows how the SAT is overestimated in the knees, pelvisand thorax, i.e. in slices containing relatively largeamounts of BM. The independence between the absolute

difference and total SAT volume measured by CTstrengthens the hypothesis that the inclusion of BM,which probably shows a limited variation with differingamounts of SAT, causes the overestimation. Ideally, BMshould be excluded from the MRI whole-body volumes.However, accurate exclusion of BM from the whole-bodyregion was found to be difficult at the current MRIresolution.

VAT was seen to be underestimated by MRI comparedwith CT. Absolute differences in VAT slice area andwhole-body volume were seen to increase with increas-ing VAT areas and volumes measured by CT, respec-tively. However, the relative differences showed no

dependence on VAT areas or volumes measured by CT.A previous study has shown that the intestinal contentmight contribute significantly to the VAT area measured

Figure 3. Whole-body subcutaneous adipose tissue (SAT)volume differences (MRI CT) plotted as a function of CTvolume measured (the reference). Differences (mean standard deviation) are illustrated using lines. No lineardependence was found.

Figure 4. Whole-body visceral adipose tissue (VAT) volumedifferences (MRI CT) plotted as a function of CT volumemeasured (the reference). The linear dependence was foundto be: 0.209CT +0.310; r250.72;p50.002.



8/14/2019 123.full.pdf

6/8

using single-slice CT analysis [18]. Intestinal contentcould possibly affect the results from the separatemodalities differently. The presence of air is, for

example, well known to cause T2* relaxation, resultingin a reduced MR signal.

Differences observed between MRI and CT wereprobably caused by errors resulting from soft-tissue

motion, motion artefacts, the MRI protocol used, partialvolume effects, differences in slice positions, imageresolution differences, image artefacts, FOV exceedingsand non-identical positioning of the subjects. Partialvolume effects probably affect the CT and MRI methodsdifferently, especially as the out-of-phase MRI protocolwas used. The fact that half bodies were acquired andregistered in the post-processing in both the CT and MRIprotocols limits the accuracy of both methods. Thepresence of dental metal in one subject gave imagingartefacts on MRI and the fact that six subjects exceededthe FOV during CT acquisition also limits the accuracy ofthe results. The previous validation of the CT protocol

was based on a comparison with total body weightmeasured by scales, and did not validate the ATmeasurements specifically. Minor differences in AT

Table 3. Total fat weights in the 10 subjects, measured byCT, MRI and DEXA

Whole-body fat weight (kg)

Subject CT MRI DEXA

1 32.78 30.90 26.072 48.43 50.06 39.093 33.95 33.84 28.424 23.29 23.24 19.775 39.94 39.88 35.966 42.05 40.27 37.277 25.24 24.35 21.198 30.58 29.18 26.339 25.35 25.49 20.52

10 58.33 57.10 53.03

DEXA, dual energy X-ray absorptiometry.

Figure 5. (a) Slice-wise mean totalbody, subcutaneous adipose tissue(SAT) and visceral adipose tissue(VAT) areas from the CT and theMRI method are displayed. The 30-slice positions used are illustrated in(b). Slice-wise total area, SAT andVAT differences (MRI CT) are givenas mean confidence interval in (c),(d) and (e), respectively. In (d), themean SAT values, after exclusion ofbone marrow (BM), are given forthe eight slices (dotted line).



8/14/2019 123.full.pdf

7/8

measurements between MRI and CT could therefore becaused by inaccuracies in either of these two methods.Multiple sources of the differences/errors mentionedabove affect both MRI and CT and introduce bias to bothmethods. This should be kept in mind when the CTresults are used as a reference.

The Flip80 data was occasionally seen to have voxels

in the skinbackground interface with values lower thanTbody (see Appendix A), e.g. by the mouth or by theknees, where the SAT layers are absent or thin. Theregion-growing of the background was seen to enterthe body at these locations. Therefore, the regiongrowing was limited by use of the binary mask Mmin.The binary mask needed to fill a volume slightly smallerthan the body. The mask created using slice-wise dilationand erosion gave an accurate border, and the maskcreated using dilation and erosion in three dimensionswas used to fill the body, as this was not always the caseusing the slice-wise approach in, for example, slices atthe level of the mouth or nose. The segmented lungswere occasionally, by visual inspection, seen to under-

estimate the lung volume. This was caused by motionartefacts from the heart and by the segmentation methodused.

Fat weight was underestimated by DEXA analysis.This is partly explained by the three heaviest subjectsexceeding the DEXA FOV. However, the underestima-tion was not eliminated by excluding these subjects,suggesting that there were additional causes of thedifferences. Large obese subjects (.120 kg) have pre-viously been reported to give deviating results [19, 20].Inter- and intra-manufacturer variations in DEXA mea-surements have also previously been reported [12, 21,22]; indeed, inter-manufacturer variations of up to 10%

of fat mass have been found [22]. Measurements ofabdominal fat have previously been shown to under-estimate the fat mass by 10% compared with CT [23].When comparing AT volume results from imagingtechniques with the fat weights estimated by DEXA, anassumption on AT fat density is needed. This densityassumption will affect directly the comparison. MRI andCT, as opposed to DEXA, allow direct separation of VATand SAT from the acquired images. These modalities aretherefore often chosen in studies of different adiposetissue depots.

One limitation of this study is the small number ofsubjects included. The body weight and body massindex range were therefore not optimal. A shortcoming

of MRI, and sometimes also of CT and DEXA, is thehardware, which limits the size and weight of thesubjects to be examined. The acquisition and post-processing required for the MRI method might also limitthe availability of this method. In the comparison of totalsubject weights (Table 2), MRI was not included, as theestimation of the non-AT class density is outside thescope of this study.

In conclusion, strong correlations between bodycomposition analysis results in whole-body regions fromCT, MRI and DEXA have been shown. We believe thatthe differences between the results from MRI vs CT areacceptable. Additionally, the MRI method allows a fully

automated post-processing of whole-body MR data,allowing longitudinal whole-body studies that are alsoapplicable to children and adolescents.

Acknowledgments

This study was supported by the Swedish ResearchCouncil, Grant no. K2006-71X-06676-24-3, and theGoteborg Medical Society.

References

1. Van Gaal LF, Mertens IL, De Block CE. Mechanisms linkingobesity with cardiovascular disease. Nature2006;444:87580.

2. Chowdhury B, Sjostrom L, Alpsten M, Kostanty J, Kvist H,Lofgren R. A multicompartment body composition techni-que based on computerized tomography. Int J Obes RelatMetab Disord 1994;18:21934.

3. Ross R. Advances in the application of imaging methods inapplied and clinical physiology. Acta Diabetol2003;40:4550.

4. Pietrobelli A, Formica C, Wang Z, Heymsfield SB. Dual-energy X-ray absorptiometry body composition model:review of physical concepts. Am J Physiol 1996;271:94151.

5. Bachrach LK. Dual energy X-ray absorptiometry (DEXA)measurements of bone density and body composition:

promise and pitfalls. J Pediatr Endocrinol Metab2000;13:9838.6. Mattsson S, Thomas BJ. Development of methods for body

composition studies. Phys Med Biol 2006;51:20328.7. Ross R, Goodpaster B, Kelley D, Boada F. Magnetic

resonance imaging in human body composition research.From quantitative to qualitative tissue measurement.Ann N Y Acad Sci 2000;904:127.

8. Ellis KJ. Human body composition:in vivomethods. PhysiolRev 2000;80:64980.

9. Kullberg J, Ahlstrom H, Johansson L, Frimmel H.Automated and reproducible segmentation of visceral andsubcutaneous adipose tissue from abdominal MRI. IntJ Obes(Lond) 2007;31:180617.

10. Formica D, Silvestri S. Biological effects of exposure to

magnetic resonance imaging: an overview. Biomed EngOnline 2004;3:11.11. Kullberg J, Angelhed JE, Lonn L, Brandberg J, Ahlstrom H,

Frimmel H, et al. Whole-body T1 mapping improves thedefinition of adipose tissue: consequences for automatedimage analysis. J Magn Reson Imaging 2006;24:394401.

12. Lantz H, Samuelson G, Bratteby LE, Mallmin H, Sjostrom L.Differences in whole body measurements by DXA-scanningusing two Lunar DPX-L machines. Int J Obes Relat MetabDisord 1999;23:76470.

13. Sjostrom L, Larsson B, Backman L, Bengtsson C, BouchardC, Dahlgren S, et al. Swedish obese subjects (SOS).Recruitment for an intervention study and a selecteddescription of the obese state. Int J Obes Relat MetabDisord 1992;16:46579.

14. Sjostrom L, Lindroos AK, Peltonen M, Torgerson J,Bouchard C, Carlsson B, et al. Lifestyle, diabetes, andcardiovascular risk factors 10 years after bariatric surgery.N Engl J Med 2004;351:268393.

15. Starck G, Lonn L, Cederblad A, Forssell-Aronsson E,Sjostrom L, Alpsten M. A method to obtain the same levelsof CT image noise for patients of various sizes, to minimizeradiation dose. Br J Radiol 2002;75:14050.

16. Nelder JA, Mead R. A simplex method for functionminimization. Comput J 1965;7:30813.

17. Rasband WS. ImageJ (version vl.37s). 1.37s ed. Bethesda,MD: National Institute of Health; 2006. [Available from:http://rsb.info.nih.gov/ij/].

18. Potretzke AM, Schmitz KH, Jensen MD. Preventing over-estimation of pixels in computed tomography assessment of

visceral fat. Obes Res 2004;12:1698701.19. Svendsen OL. Body composition and fat distribution by

dual energy X-ray absorptiometry in overweight postme-



8/14/2019 123.full.pdf

8/8

nopausal women. Effect of energy-restriction and exercise.Dan Med Bull 1996;43:24962.

20. Laskey MA. Dual-energy X-ray absorptiometry and bodycomposition. Nutrition 1996;12:4551.

21. Genton L, Hans D, Kyle UG, Pichard C. Dual-energy X-rayabsorptiometry and body composition: differences betweendevices and comparison with reference methods. Nutrition2002;18:6670.

22. Kistorp CN, Svendsen OL. Body composition analysis bydual energy X-ray absorptiometry in female diabetics differ

between manufacturers. Eur J Clin Nutr 1997;51:44954.23. Snijder MB, Visser M, Dekker JM, Seidell JC, Fuerst T,

Tylavsky F, et al. The prediction of visceral fat by dual-energy X-ray absorptiometry in the elderly: a comparisonwith computed tomography and anthropometry. Int J ObesRelat Metab Disord 2002;26:98493.

Appendix A

Segmentation of bodies and lungs from the MRIFlip80 data

The automated segmentation of the bodies used aregion-growing approach combined with two binarywhole-body masks, denoted Mcrude and Mmin. Mcrudewas used to remove motion artefacts outside the bodyand Mminwas used to prevent the region-growing ofthe background from penetrating the body. The segmen-tation of the lungs was performed by the use ofthresholding, morphological operations and filtering

based on object size.Mcrude was created by applying a region-growing in

three dimensions of the body voxels of intensity valuegreater than a threshold value (Tcrude 5 130). The resultfrom the region-growing is a binary body object contain-

ing holes from low-intensity regions inside the body. Oneiteration of morphological dilation (18 neighbourhood(NB)) was applied to close the surface of the body. Aregion-growing of the background was applied to create a

binary body object without holes in the low-intensity bodyregions. One iteration of erosion (18 NB) was applied tocorrect for the previously performed dilation.

Mminwas created from the union of two binary masks.The first binary mask was created by applying threeiterations of erosion (18 NB) in three dimensions to theMcrude mask. The three iterations were determinedempirically to result in a binary body mask enclosedwithin the subject bodies. The second binary mask was

created by region-growing of the body voxels of intensityvalue greater thanTcrude. The majority of the holes in the

body surface were closed by two iterations of dilation (4NB, axial slices) in two dimensions. A region-growing ofthe background was applied to separate the backgroundfrom the body. Three iterations of erosion (4 NB, axialslices) in two dimensions of the binary body wereapplied to ensure that the mask was slightly smaller thanthe body.

Because image intensity levels were seen to varybetween the acquired stacks, the bodies were segmentedusing a stack-wise approach. A threshold level, denotedTbody, was automatically determined for each stack. Anintensity-based region-growing of the background untilTbodywas reached was used to determine the extensionof the body in the skinbackground interface. Theregion-growing was limited to the region outside Mmin,and motion artefacts outside the body caused bypulsatile flow and motion were removed by maskingusing Mcrude.

Tbody was determined as the average of the twointensity levels IBG and IAT derived from the intensityvalues of the background and AT histogram peaks,respectively. Least square fitting of Gaussian functions tothe peaks was performed. The Gaussian function centreswere used as the intensity levels. The underlyingassumption was that voxels containing equal amountsof background and AT have an intensity value equal toTbody.

The segmentation of the lungs first used a threshold-ing of voxels in the intensity interval 080 AU inside thesegmented body, resulting in a binary volume mainlycontaining voxels originating from the lungs, intestinalgas and cortical bone. One iteration of erosion (18 NB)was applied to remove most of the unwanted binaryobjects. The binary objects were then filtered based on

their three-dimensional (18 NB) connected volumes.Objects with volumes smaller than 1000 voxels(

123.full.pdf

Documents