Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Semi-automated segmentation and quantification of adipose tissue in calf and thigh by MRI: a preliminary study in patients with monogenic metabolic syndrome

Semi-automated segmentation and quantification of adipose tissue in calf and thigh by MRI: a... Background: With the growing prevalence of obesity and metabolic syndrome, reliable quantitative imaging methods for adipose tissue are required. Monogenic forms of the metabolic syndrome include Dunnigan-variety familial partial lipodystrophy subtypes 2 and 3 (FPLD2 and FPLD3), which are characterized by the loss of subcutaneous fat in the extremities. Through magnetic resonance imaging (MRI) of FPLD patients, we have developed a method of quantifying the core FPLD anthropometric phenotype, namely adipose tissue in the mid-calf and mid-thigh regions. Methods: Four female subjects, including an FPLD2 subject (LMNA R482Q), an FPLD3 subject (PPARG F388L), and two control subjects were selected for MRI and analysis. MRI scans of subjects were performed on a 1.5T GE MR Medical system, with 17 transaxial slices comprising a 51 mm section obtained in both the mid-calf and mid-thigh regions. Using ImageJ 1.34 n software, analysis of raw MR images involved the creation of a connectedness map of the subcutaneous adipose tissue contours within the lower limb segment from a user-defined seed point. Quantification of the adipose tissue was then obtained after thresholding the connected map and counting the voxels (volumetric pixels) present within the specified region. Results: MR images revealed significant differences in the amounts of subcutaneous adipose tissue in lower limb segments of FPLD3 and FPLD2 subjects: respectively, mid-calf, 15.5% and 0%, and mid-thigh, 25.0% and 13.3%. In comparison, old and young healthy controls had values, respectively, of mid-calf, 32.5% and 26.2%, and mid-thigh, 52.2% and 36.1%. The FPLD2 patient had significantly reduced subcutaneous adipose tissue compared to FPLD3 patient. Conclusion: Thus, semi-automated quantification of adipose tissue of the lower extremity can detect differences between individuals of various lipodystrophy genotypes and represents a potentially useful tool for extended quantitative phenotypic analysis of other genetic metabolic disorders. Page 1 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Background Methods The metabolic syndrome (MetS) related to a pattern of Study subjects All study subjects were female. The study sample included central or abdominal obesity is a major health concern in the westernized world. One approach to begin to under- an FPLD2 subject (designated GL0096) and an FPLD3 stand a common complex trait such as MetS is to closely subject (designated GL0658), a young control subject study individuals who have a rare monogenic analogue of (designated GL2784) whose body mass index (BMI) was the condition. In the case of MetS, the familial partial matched to the FPLD2 subject and an older control sub- lipodystrophy syndromes represent an extreme mono- ject (designated GL2990) whose age was matched with genic model system that demonstrates the salient clinical both FPLD subjects and whose BMI was matched to the (increased blood pressure and increased abdominal fat) FPLD3 subject (Table 1). All subjects provided informed and biochemical manifestations (increased plasma glu- consent to participate and human ethics approval was cose and triglyceride concentrations and decreased obtained from the University of Western Ontario Institu- plasma HDL cholesterol concentration). tional Review Board (protocol #11244). Clinical and biochemical assessment The two molecular forms of autosomal dominant Dunni- gan-type familial partial lipodystrophy (FPLD) result All subjects provided a medical history and were subjected from mutations either in LMNA encoding nuclear lamin to a complete physical examination. Bioimpedance anal- A/C (FPLD2; MIM 151660) or in PPARG encoding perox- ysis (BIA) measurements were also gathered using the isome proliferator-activated receptor-γ (FPLD3; MIM Tanita BC-418 Segmental Body Composition Analyzer 604367) [1-3]. One in 100,000 individuals has FPLD in (Tanita, Arlington Heights, IL) providing estimates of per- North America. Patients with either form of this rare dis- cent fat for the total body and lower right and left extrem- order show loss of subcutaneous fat, especially from ities. The average of three measurements was reported for extremities, together with predisposition to insulin-resist- each BIA value. ant diabetes, dyslipidemia and hypertension [1-3]. Despite the similar clinical course, there are subtle clinical Magnetic resonance imaging and image analysis phenotypic differences between FPLD2 and FPLD3 [1-3]. MRI scans were obtained at the London Health Sciences Centre, University Campus, London, Ontario. Scanning For instance, compared to FPLD2 subjects, FPLD3 sub- jects appear to have less severe adipose involvement on was performed on a 1.5T GE MR Medical system (Model: physical examination, together with more severe clinical Signa Excite) using an 8-channel receive-only torso array and biochemical manifestations of insulin resistance, and coil. Images of the various sections were acquired using a more variable response to treatment with thiazolidinedi- T1-weighted Spin Echo pulse sequence with the following one drugs [2,3]. parameters: FOV of 40 cm for mid-calves and 48 cm for mid-thighs, TR/TE 400/10 ms, bandwidth +/-15.63 kHz, 2 To date, thorough semi-quantitative descriptions of the NEX (number of signal averages), and an acquisition localization and extent of fat loss from affected tissues matrix of 256 × 256. have taken advantage of both clinical assessment and, more recently, non-invasive imaging methods, such as Mid-calf and mid-thigh sections were positioned based on magnetic resonance imaging (MRI) [4-6]. While the reference anatomical features. The mid-point of the tibia descriptions of MR images in FPLD patients have been was selected for "mid-calf" measurements and the mid- extensive, thorough and detailed, they have not yet been point of the femur was selected for "mid-thigh" measure- quantitative [4-6]. Because quantitation of fat mass on ments (Figure 1). For both the mid-calf and mid-thigh MRI could: 1) enhance the description of these rare disor- regions we acquired image stacks comprised of 17 transax- ders; 2) allow for statistical comparisons; and 3) yield new ial slice images of 3 mm thickness each, together compris- quantitative traits to follow serially, it is important to ing a total superior/inferior coverage of 51 mm. The MRI develop robust and replicable tools and methods to quan- provided a bright, high-threshold adipose tissue signal in tify subcutaneous fat [7,8]. We now report a method to raw image data, relative to other tissues and background. quantify lower extremity fat depots in patients with Other tissues, such as muscle and connective tissue, FPLD2 and FPLD3 from serial MR images that utilized an appeared as dark regions with low threshold values in MR almost completely automated strategy. Using the method images (Figure 2). The whole body scans (Figure 1) are a developed, the study further reports on quantitative differ- composite of four mid-slice coronal stack images acquired ences between lower extremity adipose tissue distribution at four stations (head/neck, thoracic/abdominal, pelvic/ in the case of two FPLD patients along with comparisons thigh, lower leg). to matched controls. Page 2 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Table 1: Characteristics of FPLD patients compared to controls GL2784 GL2990 GL0658 GL0096 Diagnosis Control Control FPLD3 FPLD2 Mutation Wild-type Wild-type PPARG F388L LMNA R482Q Age (years) 24 50 49 63 Sex Female Female Female Female Height (m) 1.63 1.60 1.52 1.53 Weight (kg) 61.7 89.1 80.2 58.1 Body mass index (kg/m ) 23.5 34.8 33.4 24.8 Waist circumference (cm) 78.8 103.6 105.7 88.3 Waist-to-hip circumference ratio 0.88 0.86 0.88 0.92 Blood pressure (mmHg) 113/63 154/89 138/88 (treated) 131/76 (treated) BIA measures (PBF, %) total body 28.0 ± 0.2 47.5 ± 0.2 31.8 ± 0.2 29.7 ± 0.1 right leg 31.4 ± 0.1 49.5 ± 0.1 44.0 ± 0.2 36.8 ± 0.1 left leg 31.3 ± 0.1 49.2 ± 0.1 47.9 ± 0.5 37.7 ± 0.1 Mean sc+inf volume/slice (%) mid-calf 26.3 ± 1.1 34.8 ± 1.5 19.2 ± 1.7 N/A mid-thigh 44.3 ± 2.1 56.1 ± 1.5 34.4 ± 2.5 24.3 ± 3.7 Overall sc+inf volume (%) mid-calf 26.2 34.7 19.2 N/A mid-thigh 44.5 56.1 34.5 24.5 Abbreviations: FPLD, familial partial lipodystrophy; BIA, bioimpedance analysis; PBF, percent body fat; MRI, magnetic resonance imaging; sc+inf fat, subcutaneous plus infiltrated fat; N/A, could not be assessed (no visible subcutaneous fat) Fat quantification method from MRI pose tissue threshold value ranges were obtained by man- Analysis of the MRI stack images and measurements of ually sampling the signal intensity in each image stack. subcutaneous adipose tissue was done by a single Using this threshold selection mechanism, the connected- observer using analysis protocols developed in our labo- ness map of subcutaneous adipose tissue was then thresh- ratories (Figure 3). For each MRI data set acquired, the olded to segment the volume to be analyzed. Finally, the subcutaneous adipose tissue volume was quantified using contours of segmented subcutaneous adipose tissue were ImageJ version 1.34 n image analysis software [9], specif- quantitated using the Voxel Counter tool, which pro- ically utilizing the Connected Threshold Grower and duced the final output voxel count within the volumetric Voxel Counter tools. Subcutaneous adipose tissue was region. defined as the adipose tissue that circulated the circumfer- ence of the lower limb, adjacent to the skin, as well as any Percent adipose tissue was calculated by dividing the total connected adipose tissue that was infiltrated into the mus- voxels determined for fat intensity signals connected to cle. the subcutaneous adipose seed point by the total voxels for the slice (Figure 3). After the percent adipose tissue was Prior to image analysis and fat quantification, all raw calculated for each slice, the average of 17 slices was images underwent a preprocessing stage using the auto- reported as the average percent adipose tissue/slice. The brightness tool in order to minimize background noise percent volume of fat for each region was also obtained by and improve the quality of the images as much as possi- adding the fat intensity voxels for all 17 slices (total fat ble. Images were further standardized with a distance in volume) and then dividing them by the sum of the total pixels set at 1.00 pixel/mm and the image dynamic was voxel areas (total region volume). reduced to an 8-bit type to match the requirements of the used software. Starting from a user-defined seed point Statistical analysis within the subcutaneous adipose tissue in the image, the Subcutaneous adipose measurements from MRI were sta- method utilized the Connected Threshold Grower tool to tistically analyzed using SAS version 8.2 (SAS Institute, create a connectedness map of the volumetric contours of Cary, NC). The Pearson correlation coefficient was used to subcutaneous adipose tissue within the image stack. This test variation among duplicate blinded subcutaneous adi- map represented the strength of connectivity between the pose tissue measurements made by the same observer on seed point in the subcutaneous region and every voxel different days in the mid-calf and mid-thigh (intra- (volumetric pixel) in the image stack. Total tissue and adi- observer variability) and also to test inter-observer varia- Page 3 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 F th Figure 1 uigh ll body coronal magnetic resonance image of the subjects in the study and the assigned survey fields in the mid-calf and mid- Full body coronal magnetic resonance image of the subjects in the study and the assigned survey fields in the mid-calf and mid-thigh. On these survey images the horizontal bars indicate the location of the mid-calf and mid-thigh sec- tions, positioned based on reference anatomical features. The mid-point of the tibia was selected for "mid-calf" measurements and the mid-point of the femur was selected for "mid-thigh" measurements. Subjects from left to right are normal controls, GL2784 and GL2990, followed by the FPLD3 patient (GL0658) and FPLD2 patient (GL0096). tion for the same samples, analyzed by a different tion Program Adult Treatment Program III (NCEP ATP III) observer. The t-test for unequal variances was used to test criteria for diagnosis of the metabolic syndrome (MetS), for differences in mean percent subcutaneous adipose had higher BMI, waist circumference (but not ratio of between FPLD subjects and normal controls. A nominal waist-to-hip circumference), and percent body fat (PBF) P-value < 0.05 was chosen as the threshold for significance for the total body and lower extremities, as measured by for all statistical comparisons. segmental body composition MRI analysis. Compared to controls, FPLD3 subject GL0658, who also met the blood pressure and waist circumference cut points of the NCEP Results Baseline clinical and anthropometric features of study ATP III criteria for MetS, had similar ratio of waist-to-hip subjects circumference, similar BMI and percent lower extremity The clinical and anthropometric features of the study sub- fat to age-matched control GL2990, but lower total body jects are displayed in Table 1. Compared to control subject PBF compared to the same age-matched control (Table 1). GL2784 (female, aged 24), control subject GL2990 Compared to controls, FPLD2 subject GL0096, who also (female, aged 50), who met blood pressure and waist cir- met blood pressure and waist circumference criteria of the cumference cut points of the National Cholesterol Educa- NCEP ATP III definition for MetS, had increased ratio of Page 4 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Tr jects in the study Figure 2 ansaxial magnetic resonance images at the levels of mid-calf (top slice images) and mid-thigh (bottom slice images) of the sub- Transaxial magnetic resonance images at the levels of mid-calf (top slice images) and mid-thigh (bottom slice images) of the subjects in the study. Bright/white signals in these images are highlighting adipose tissue within these ana- tomical sections. Dark signals represent either muscle tissue within sections or the background of the images. Subject GL2784 is a healthy 24 year old woman whose MRI showed no infiltrated fat into calf muscle, and only small amount of infiltration in the thigh. Subject GL2990 is a normal 50 year old woman who had somewhat increased subcutaneous (sc) fat in the calves and mid-thigh with slightly more infiltration of fat into the muscle compared to the images of subject GL2784. Subject GL0658 is a 49 year old FPLD3 patient (heterozygous for mutation PPARG F388L) whose scans show moderate loss of sc fat in both the calves and mid thigh and moderate levels of fat infiltration. Subject GL0096 is a 63 year old FPLD2 patient (heterozygous for mutation LMNA R482Q) whose scan shows total sc fat loss in the calves, major sc fat loss in the mid-thigh and marbled appear- ance of muscle tissue due to severe amounts of fat being stored within the muscle. waist-to-hip circumference (android pattern). BMI and right and left sides. The overall inter-observer correlation PBF for the total body and lower extremities for FPLD2 coefficients were, on average, 0.988 for the mid-calf and subject GL0096 was similar to young control GL2784 and 0.991 for the mid-thigh. significantly less than older control GL2990 (Table 1). Quantification of subcutaneous fat from MRI Qualitative differences on survey MRIs Quantification of the percent subcutaneous adipose tissue Qualitative coronal regional fat distribution profile differ- present in the mid-calf and mid-thigh regions showed that ences between affected and normal controls are shown in the control subjects had values ranging from 26-56%, Figure 1. The main visible differences included: 1) greater with mid-thigh values always greater than mid-calf values. subcutaneous fat depots, especially around the hips and The older control subject GL2990 (BMI 34.8) had percent thighs, for the two control subjects compared with the subcutaneous adipose tissue values that were ~1.3-fold FPLD3 and FPLD2 subjects; and 2) attenuation of subcu- greater than the younger, normal weight control subject taneous fat stores at a lower point on the thigh of the GL2784 P < 0.0001). The FPLD3 subject GL0658 had sig- FPLD3 subject compared to the FPLD2 subject. nificantly lower percent adipose tissue values for both the mid-calf and mid-thigh regions in comparison to both Intra- and inter-observer correlations for quantitative MRI control subjects (P < 0.0001 for both). The most signifi- analysis cant attenuation in subcutaneous adipose tissue was Intra-observer correlation was determined by comparing observed for the FPLD2 subject, where no subcutaneous two replicates of percent subcutaneous fat for both the connectedness map of fat was attainable for the minute mid-calf and mid-thigh derived from subjects GL2784, remnants of adipose tissue present in the perimeter of GL2990, GL0658 and GL0096. Each replicate involved mid-calf region, and thus quantification using the auto- analysis of 17 transaxial images for both the right and left mated Connected Threshold Grower tool was impossible. sides. Intra-observer correlation coefficients based on at The percent subcutaneous adipose tissue in the mid-thigh least 68 sections each were, on average, 0.996 for the mid- of FPLD2 subject GL0096 was also significantly lower calf and 0.998 for the mid-thigh. Inter-observer correla- than that observed for FPLD3 subject GL0658 (24.3 ± tion was determined by comparing percent subcutaneous 3.7% vs 34.4 ± 2.5%, P < 0.0001). fat for both the mid-calf and mid-thigh derived from sub- jects GL2784, GL2990, GL0658 and GL0096, as measured Mean and overall percent adipose tissue values are by two independent observers. Each determination reported (Table 1), representing averaged values of indi- involved analysis of 17 transaxial images for both the vidual slices composing each respective image stack and Page 5 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 sue in the thigh was 24.3 ± 3.7% and 34.4 ± 2.5% (P < 0.0001), for the FPLD2 and FPLD3 patients respectively. Current clinical assessment of adipose tissue distribution in common obesity and metabolic syndrome and subjects with FPLD2 and FPLD3 is still in its infancy. Also, BIA failed to capture differences in percent fat in lower extrem- ities in FPLD2 vs FPLD3 perhaps because so much fat was infiltrated into muscle in FPLD2. In contrast, MRI adipose connectedness maps and semi-automated subcutaneous adipose tissue quantification with very high resolution and reproducibility, captured traits that could be com- pared statistically, confirming the subtle clinical differ- ences [3,10]. This semi-automated method involved a Connected Threshold Grower tool which specified inclusion of only adipose tissue connected to the initial subcutaneous seed point. Based on this pilot study of FPLD patients, we observed very high intra- and inter-observer correlation values: r > 0.99 and >0.98, respectively. In addition to its reproducibility, the described method yields results quickly and accurately, with minimal user intervention. The method was limited by including only connected Quan Figure 3 tification of percent adipose tissue infiltrated adipose tissue. However, given the imprecise Quantification of percent adipose tissue. For each of definition of subcutaneous adipose tissue in extremities, the 17 transaxial slices in a given anatomical section, both the we elected to include the connected infiltrated adipose tis- total volume and the total subcutaneous (sc) and connected sue in our calculations, again since this would require no infiltrated (inf) fat volumes were selected using the Con- user judgment and/or intervention, thus reducing another nected Threshold Grower tool. Their corresponding vol- potential source of analytic variation. An additional limi- umes were determined using the Voxel Counter tool. The tation inherent in the ImageJ software, which does not percent adipose tissue was calculated for each slice by divid- affect reproducibility but affects image dynamic, is that of ing the total voxels determined for the sc + inf fat by the the 16-bit to 8-bit change to the image stacks prior to anal- total voxels for the slice. The percent adipose tissue was determined for each slice alone and also for the overall sec- ysis. This reduction in image dynamic, which reduces res- tion, combining the results from all 17 slices. olution, is a common setback in medical image processing where similar general-purpose software librar- ies are used. Future development of the software to utilize values of total volume subcutaneous adipose tissue original raw images would be advantageous in maintain- respectively. Each of these values is also an average of two ing image integrity and reflecting more accurate analysis replicate data sets from two independent analyses. The data acquired from quantification. correlation (r) between mean subcutaneous fat areas and overall fat volume was 0.99998. Evaluating FPLD patients theoretically allowed for assess- ment of the lower limits of resolution of the method; however, the method appeared insensitive for calf adipose Discussion Using a strategy to quantify subcutaneous fat in the lower measurements in FPLD2, since there was no subcutane- extremity that was based on connectivity analysis, we ous fat according to the definition specified in the quanti- found significant differences between subcutaneous adi- fication methodology. Future application of this pose tissue in the mid-calf and mid-thigh sections of FPLD quantification method may include quantification of patients compared to normal controls. We found signifi- both thigh and calf depots for "garden variety" obesity, cantly reduced lower extremity subcutaneous adipose tis- metabolic syndrome or diabetes. This approach might sue in a subject with FPLD2 than in a subject with FPLD3. also be applicable to quantify metabolically important Specifically, no subcutaneous adipose tissue could be substrata of fat [11]. quantified in the calf of the FPLD2 patient compared to 19.2 ± 1.7% subcutaneous adipose tissue in FPLD3 (P < We recognize that this study was limited due to the small 0.0001). Similarly, the percent subcutaneous adipose tis- subject numbers from whom subcutaneous adipose tissue Page 6 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 values were extracted. Acquisition of such values from a Competing interests larger number of patients with both FPLD subtypes would The author(s) declare that they have no competing inter- verify the likely results observed here. Furthermore, con- ests. trols were not ideally matched for age and BMI: while the FPLD2 patient had a similar BMI as the young control Authors' contributions individual, unmeasured and uncontrolled factors related SAA participated in the experimental design, data acquisi- to age might have further contributed to variation in sub- tion and analysis, interpretation of results, and manu- cutaneous adipose tissue. Expanding the sample size in script writing. RLP participated in the analysis of the MRI future studies would clearly be helpful in this regard. data and manuscript writing. JFR participated in data acquisition, analysis and interpretation of results. BAM The whole body scans suggested that this method can be was involved in the clinical assessment. RW performed adapted for other fat depots or bodily organs. However, the MRI scans. BKR participated in the experimental widespread application would depend on development of design, data analysis, and interpretation of results. RAH standards with respect to regions surveyed, anatomical participated in the experimental design, data analysis, landmarks, number of measurements, etc – similar to the interpretation of results and manuscript writing. All consensus standards agreed upon for carotid intima- authors approved the final manuscript. media thickness measurements using ultrasound. Also, intramuscular fat is distributed either in intra- or inter- Acknowledgements Supported by the Jacob J. Wolfe Distinguished Medical Research Chair, the myocellular depots; which could be more specifically Edith Schulich Vinet Canada Research Chair (Tier I) in Human Genetics, a evaluated using proton magnetic resonance spectroscopy Career Investigator award from the Heart and Stroke Foundation of (MRS) and/or fat selective MRI [12-14]. Such regional dis- Ontario, and operating grants from the Canadian Institutes for Health tribution could be an additional MRI analyte that could Research, the Heart and Stroke Foundation of Ontario (NA5320), the be considered together with other intermediate traits in Ontario Research and Development Challenge Fund (Project #0507) and subjects with FPLD or even common metabolic syn- by Genome Canada through the Ontario Genomics Institute. drome. Furthermore, it is possible to obtain carbon-13 nuclear magnetic resonance (NMR) spectra of human References muscle glycogen in vivo in diabetic patients [15], which 1. Garg A: Acquired and inherited lipodystrophies. N Engl J Med 2004, 350(12):1220-1234. has helped understand the pathogenesis of insulin resist- 2. Hegele RA: Phenomics, lipodystrophy, and the metabolic syn- ance, metabolic syndrome and type 2 diabetes. Quantifi- drome. Trends Cardiovasc Med 2004, 14(4):133-137. 3. Hegele RA: Lessons from human mutations in PPARgamma. cation of fat depots using MRI and appropriate image Int J Obes (Lond) 2005, 29 Suppl 1:S31-5. analysis software could provide complementary analytes 4. Agarwal AK, Garg A: A novel heterozygous mutation in perox- for research and perhaps eventually for the diagnosis and isome proliferator-activated receptor-gamma gene in a patient with familial partial lipodystrophy. J Clin Endocrinol monitoring of interventions. Metab 2002, 87(1):408-411. 5. Garg A, Peshock RM, Fleckenstein JL: Adipose tissue distribution pattern in patients with familial partial lipodystrophy (Dun- Conclusion nigan variety). J Clin Endocrinol Metab 1999, 84(1):170-174. In summary, we report the use of MRI and image analysis 6. Garg A, Vinaitheerthan M, Weatherall PT, Bowcock AM: Pheno- software employing Connected Threshold Grower and typic heterogeneity in patients with familial partial lipodys- trophy (dunnigan variety) related to the site of missense Voxel Counter tools to help quantify lower extremity sub- mutations in lamin a/c gene. J Clin Endocrinol Metab 2001, cutaneous fat depots in patients with two molecular forms 86(1):59-65. of partial lipodystrophy. We also showed that the meas- 7. Iacobellis G: Imaging of visceral adipose tissue: an emerging diagnostic tool and therapeutic target. Curr Drug Targets Cardi- urements showed high intra- and inter-observer correla- ovasc Haematol Disord 2005, 5(4):345-353. tion in a small sample. Finally, the measurements could 8. Liou TH, Chan WP, Pan LC, Lin PW, Chou P, Chen CH: Fully auto- mated large-scale assessment of visceral and subcutaneous be compared statistically and thus confirmed the clinical abdominal adipose tissue by magnetic resonance imaging. impression that FPLD2 and FPLD3 differ with respect to Int J Obes (Lond) 2006, 30(5):844-852. the extent of subcutaneous fat loss; specifically, subcuta- 9. ImageJ: Image processing and analysis in Java [http:// rsb.info.nih.gov/ij/] neous fat loss in the FPLD2 subject is greater than in the 10. Hegele RA, Cao H, Frankowski C, Mathews ST, Leff T: PPARG FPLD3 individual. Increasing the sample size of FPLD F388L, a transactivation-deficient mutant, in familial partial subjects in future studies will validate this interpretation. lipodystrophy. Diabetes 2002, 51(12):3586-3590. 11. Smith SR, Lovejoy JC, Greenway F, Ryan D, deJonge L, de la Bretonne These tools can be applied immediately and might be use- J, Volafova J, Bray GA: Contributions of total body fat, abdomi- ful in quantitative phenotype analysis of other forms of nal subcutaneous adipose tissue compartments, and visceral adipose tissue to the metabolic complications of obesity. lipodystrophy and in less extreme disorders of fat redistri- Metabolism 2001, 50(4):425-435. bution or repartitioning, such as "garden variety" obesity, 12. Boesch C, Slotboom J, Hoppeler H, Kreis R: In vivo determination insulin resistance, or type 2 diabetes. of intra-myocellular lipids in human muscle by means of localized 1H-MR-spectroscopy. Magn Reson Med 1997, 37(4):484-493. Page 7 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 13. Brechtel K, Jacob S, Machann J, Hauer B, Nielsen M, Meissner HP, Matthaei S, Haering HU, Claussen CD, Schick F: Acquired general- ized lipoatrophy (AGL): highly selective MR lipid imaging and localized (1)H-MRS. J Magn Reson Imaging 2000, 12(2):306-310. 14. Schick F, Eismann B, Jung WI, Bongers H, Bunse M, Lutz O: Compar- ison of localized proton NMR signals of skeletal muscle and fat tissue in vivo: two lipid compartments in muscle tissue. Magn Reson Med 1993, 29(2):158-167. 15. Petersen KF, Shulman GI: Pathogenesis of skeletal muscle insu- lin resistance in type 2 diabetes mellitus. Am J Cardiol 2002, 90(5A):11G-18G. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2342/6/11/prepub Publish with Bio Med Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical researc h in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 8 of 8 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Medical Imaging Springer Journals

Semi-automated segmentation and quantification of adipose tissue in calf and thigh by MRI: a preliminary study in patients with monogenic metabolic syndrome

Download PDF
8 pages

Loading next page...
 
/lp/springer-journals/semi-automated-segmentation-and-quantification-of-adipose-tissue-in-0ZR0DT0Uin

References (17)

Publisher
Springer Journals
Copyright
Copyright © 2006 by Al-Attar et al; licensee BioMed Central Ltd.
Subject
Medicine & Public Health; Imaging / Radiology
eISSN
1471-2342
DOI
10.1186/1471-2342-6-11
pmid
16945131
Publisher site
See Article on Publisher Site

Abstract

Background: With the growing prevalence of obesity and metabolic syndrome, reliable quantitative imaging methods for adipose tissue are required. Monogenic forms of the metabolic syndrome include Dunnigan-variety familial partial lipodystrophy subtypes 2 and 3 (FPLD2 and FPLD3), which are characterized by the loss of subcutaneous fat in the extremities. Through magnetic resonance imaging (MRI) of FPLD patients, we have developed a method of quantifying the core FPLD anthropometric phenotype, namely adipose tissue in the mid-calf and mid-thigh regions. Methods: Four female subjects, including an FPLD2 subject (LMNA R482Q), an FPLD3 subject (PPARG F388L), and two control subjects were selected for MRI and analysis. MRI scans of subjects were performed on a 1.5T GE MR Medical system, with 17 transaxial slices comprising a 51 mm section obtained in both the mid-calf and mid-thigh regions. Using ImageJ 1.34 n software, analysis of raw MR images involved the creation of a connectedness map of the subcutaneous adipose tissue contours within the lower limb segment from a user-defined seed point. Quantification of the adipose tissue was then obtained after thresholding the connected map and counting the voxels (volumetric pixels) present within the specified region. Results: MR images revealed significant differences in the amounts of subcutaneous adipose tissue in lower limb segments of FPLD3 and FPLD2 subjects: respectively, mid-calf, 15.5% and 0%, and mid-thigh, 25.0% and 13.3%. In comparison, old and young healthy controls had values, respectively, of mid-calf, 32.5% and 26.2%, and mid-thigh, 52.2% and 36.1%. The FPLD2 patient had significantly reduced subcutaneous adipose tissue compared to FPLD3 patient. Conclusion: Thus, semi-automated quantification of adipose tissue of the lower extremity can detect differences between individuals of various lipodystrophy genotypes and represents a potentially useful tool for extended quantitative phenotypic analysis of other genetic metabolic disorders. Page 1 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Background Methods The metabolic syndrome (MetS) related to a pattern of Study subjects All study subjects were female. The study sample included central or abdominal obesity is a major health concern in the westernized world. One approach to begin to under- an FPLD2 subject (designated GL0096) and an FPLD3 stand a common complex trait such as MetS is to closely subject (designated GL0658), a young control subject study individuals who have a rare monogenic analogue of (designated GL2784) whose body mass index (BMI) was the condition. In the case of MetS, the familial partial matched to the FPLD2 subject and an older control sub- lipodystrophy syndromes represent an extreme mono- ject (designated GL2990) whose age was matched with genic model system that demonstrates the salient clinical both FPLD subjects and whose BMI was matched to the (increased blood pressure and increased abdominal fat) FPLD3 subject (Table 1). All subjects provided informed and biochemical manifestations (increased plasma glu- consent to participate and human ethics approval was cose and triglyceride concentrations and decreased obtained from the University of Western Ontario Institu- plasma HDL cholesterol concentration). tional Review Board (protocol #11244). Clinical and biochemical assessment The two molecular forms of autosomal dominant Dunni- gan-type familial partial lipodystrophy (FPLD) result All subjects provided a medical history and were subjected from mutations either in LMNA encoding nuclear lamin to a complete physical examination. Bioimpedance anal- A/C (FPLD2; MIM 151660) or in PPARG encoding perox- ysis (BIA) measurements were also gathered using the isome proliferator-activated receptor-γ (FPLD3; MIM Tanita BC-418 Segmental Body Composition Analyzer 604367) [1-3]. One in 100,000 individuals has FPLD in (Tanita, Arlington Heights, IL) providing estimates of per- North America. Patients with either form of this rare dis- cent fat for the total body and lower right and left extrem- order show loss of subcutaneous fat, especially from ities. The average of three measurements was reported for extremities, together with predisposition to insulin-resist- each BIA value. ant diabetes, dyslipidemia and hypertension [1-3]. Despite the similar clinical course, there are subtle clinical Magnetic resonance imaging and image analysis phenotypic differences between FPLD2 and FPLD3 [1-3]. MRI scans were obtained at the London Health Sciences Centre, University Campus, London, Ontario. Scanning For instance, compared to FPLD2 subjects, FPLD3 sub- jects appear to have less severe adipose involvement on was performed on a 1.5T GE MR Medical system (Model: physical examination, together with more severe clinical Signa Excite) using an 8-channel receive-only torso array and biochemical manifestations of insulin resistance, and coil. Images of the various sections were acquired using a more variable response to treatment with thiazolidinedi- T1-weighted Spin Echo pulse sequence with the following one drugs [2,3]. parameters: FOV of 40 cm for mid-calves and 48 cm for mid-thighs, TR/TE 400/10 ms, bandwidth +/-15.63 kHz, 2 To date, thorough semi-quantitative descriptions of the NEX (number of signal averages), and an acquisition localization and extent of fat loss from affected tissues matrix of 256 × 256. have taken advantage of both clinical assessment and, more recently, non-invasive imaging methods, such as Mid-calf and mid-thigh sections were positioned based on magnetic resonance imaging (MRI) [4-6]. While the reference anatomical features. The mid-point of the tibia descriptions of MR images in FPLD patients have been was selected for "mid-calf" measurements and the mid- extensive, thorough and detailed, they have not yet been point of the femur was selected for "mid-thigh" measure- quantitative [4-6]. Because quantitation of fat mass on ments (Figure 1). For both the mid-calf and mid-thigh MRI could: 1) enhance the description of these rare disor- regions we acquired image stacks comprised of 17 transax- ders; 2) allow for statistical comparisons; and 3) yield new ial slice images of 3 mm thickness each, together compris- quantitative traits to follow serially, it is important to ing a total superior/inferior coverage of 51 mm. The MRI develop robust and replicable tools and methods to quan- provided a bright, high-threshold adipose tissue signal in tify subcutaneous fat [7,8]. We now report a method to raw image data, relative to other tissues and background. quantify lower extremity fat depots in patients with Other tissues, such as muscle and connective tissue, FPLD2 and FPLD3 from serial MR images that utilized an appeared as dark regions with low threshold values in MR almost completely automated strategy. Using the method images (Figure 2). The whole body scans (Figure 1) are a developed, the study further reports on quantitative differ- composite of four mid-slice coronal stack images acquired ences between lower extremity adipose tissue distribution at four stations (head/neck, thoracic/abdominal, pelvic/ in the case of two FPLD patients along with comparisons thigh, lower leg). to matched controls. Page 2 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Table 1: Characteristics of FPLD patients compared to controls GL2784 GL2990 GL0658 GL0096 Diagnosis Control Control FPLD3 FPLD2 Mutation Wild-type Wild-type PPARG F388L LMNA R482Q Age (years) 24 50 49 63 Sex Female Female Female Female Height (m) 1.63 1.60 1.52 1.53 Weight (kg) 61.7 89.1 80.2 58.1 Body mass index (kg/m ) 23.5 34.8 33.4 24.8 Waist circumference (cm) 78.8 103.6 105.7 88.3 Waist-to-hip circumference ratio 0.88 0.86 0.88 0.92 Blood pressure (mmHg) 113/63 154/89 138/88 (treated) 131/76 (treated) BIA measures (PBF, %) total body 28.0 ± 0.2 47.5 ± 0.2 31.8 ± 0.2 29.7 ± 0.1 right leg 31.4 ± 0.1 49.5 ± 0.1 44.0 ± 0.2 36.8 ± 0.1 left leg 31.3 ± 0.1 49.2 ± 0.1 47.9 ± 0.5 37.7 ± 0.1 Mean sc+inf volume/slice (%) mid-calf 26.3 ± 1.1 34.8 ± 1.5 19.2 ± 1.7 N/A mid-thigh 44.3 ± 2.1 56.1 ± 1.5 34.4 ± 2.5 24.3 ± 3.7 Overall sc+inf volume (%) mid-calf 26.2 34.7 19.2 N/A mid-thigh 44.5 56.1 34.5 24.5 Abbreviations: FPLD, familial partial lipodystrophy; BIA, bioimpedance analysis; PBF, percent body fat; MRI, magnetic resonance imaging; sc+inf fat, subcutaneous plus infiltrated fat; N/A, could not be assessed (no visible subcutaneous fat) Fat quantification method from MRI pose tissue threshold value ranges were obtained by man- Analysis of the MRI stack images and measurements of ually sampling the signal intensity in each image stack. subcutaneous adipose tissue was done by a single Using this threshold selection mechanism, the connected- observer using analysis protocols developed in our labo- ness map of subcutaneous adipose tissue was then thresh- ratories (Figure 3). For each MRI data set acquired, the olded to segment the volume to be analyzed. Finally, the subcutaneous adipose tissue volume was quantified using contours of segmented subcutaneous adipose tissue were ImageJ version 1.34 n image analysis software [9], specif- quantitated using the Voxel Counter tool, which pro- ically utilizing the Connected Threshold Grower and duced the final output voxel count within the volumetric Voxel Counter tools. Subcutaneous adipose tissue was region. defined as the adipose tissue that circulated the circumfer- ence of the lower limb, adjacent to the skin, as well as any Percent adipose tissue was calculated by dividing the total connected adipose tissue that was infiltrated into the mus- voxels determined for fat intensity signals connected to cle. the subcutaneous adipose seed point by the total voxels for the slice (Figure 3). After the percent adipose tissue was Prior to image analysis and fat quantification, all raw calculated for each slice, the average of 17 slices was images underwent a preprocessing stage using the auto- reported as the average percent adipose tissue/slice. The brightness tool in order to minimize background noise percent volume of fat for each region was also obtained by and improve the quality of the images as much as possi- adding the fat intensity voxels for all 17 slices (total fat ble. Images were further standardized with a distance in volume) and then dividing them by the sum of the total pixels set at 1.00 pixel/mm and the image dynamic was voxel areas (total region volume). reduced to an 8-bit type to match the requirements of the used software. Starting from a user-defined seed point Statistical analysis within the subcutaneous adipose tissue in the image, the Subcutaneous adipose measurements from MRI were sta- method utilized the Connected Threshold Grower tool to tistically analyzed using SAS version 8.2 (SAS Institute, create a connectedness map of the volumetric contours of Cary, NC). The Pearson correlation coefficient was used to subcutaneous adipose tissue within the image stack. This test variation among duplicate blinded subcutaneous adi- map represented the strength of connectivity between the pose tissue measurements made by the same observer on seed point in the subcutaneous region and every voxel different days in the mid-calf and mid-thigh (intra- (volumetric pixel) in the image stack. Total tissue and adi- observer variability) and also to test inter-observer varia- Page 3 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 F th Figure 1 uigh ll body coronal magnetic resonance image of the subjects in the study and the assigned survey fields in the mid-calf and mid- Full body coronal magnetic resonance image of the subjects in the study and the assigned survey fields in the mid-calf and mid-thigh. On these survey images the horizontal bars indicate the location of the mid-calf and mid-thigh sec- tions, positioned based on reference anatomical features. The mid-point of the tibia was selected for "mid-calf" measurements and the mid-point of the femur was selected for "mid-thigh" measurements. Subjects from left to right are normal controls, GL2784 and GL2990, followed by the FPLD3 patient (GL0658) and FPLD2 patient (GL0096). tion for the same samples, analyzed by a different tion Program Adult Treatment Program III (NCEP ATP III) observer. The t-test for unequal variances was used to test criteria for diagnosis of the metabolic syndrome (MetS), for differences in mean percent subcutaneous adipose had higher BMI, waist circumference (but not ratio of between FPLD subjects and normal controls. A nominal waist-to-hip circumference), and percent body fat (PBF) P-value < 0.05 was chosen as the threshold for significance for the total body and lower extremities, as measured by for all statistical comparisons. segmental body composition MRI analysis. Compared to controls, FPLD3 subject GL0658, who also met the blood pressure and waist circumference cut points of the NCEP Results Baseline clinical and anthropometric features of study ATP III criteria for MetS, had similar ratio of waist-to-hip subjects circumference, similar BMI and percent lower extremity The clinical and anthropometric features of the study sub- fat to age-matched control GL2990, but lower total body jects are displayed in Table 1. Compared to control subject PBF compared to the same age-matched control (Table 1). GL2784 (female, aged 24), control subject GL2990 Compared to controls, FPLD2 subject GL0096, who also (female, aged 50), who met blood pressure and waist cir- met blood pressure and waist circumference criteria of the cumference cut points of the National Cholesterol Educa- NCEP ATP III definition for MetS, had increased ratio of Page 4 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 Tr jects in the study Figure 2 ansaxial magnetic resonance images at the levels of mid-calf (top slice images) and mid-thigh (bottom slice images) of the sub- Transaxial magnetic resonance images at the levels of mid-calf (top slice images) and mid-thigh (bottom slice images) of the subjects in the study. Bright/white signals in these images are highlighting adipose tissue within these ana- tomical sections. Dark signals represent either muscle tissue within sections or the background of the images. Subject GL2784 is a healthy 24 year old woman whose MRI showed no infiltrated fat into calf muscle, and only small amount of infiltration in the thigh. Subject GL2990 is a normal 50 year old woman who had somewhat increased subcutaneous (sc) fat in the calves and mid-thigh with slightly more infiltration of fat into the muscle compared to the images of subject GL2784. Subject GL0658 is a 49 year old FPLD3 patient (heterozygous for mutation PPARG F388L) whose scans show moderate loss of sc fat in both the calves and mid thigh and moderate levels of fat infiltration. Subject GL0096 is a 63 year old FPLD2 patient (heterozygous for mutation LMNA R482Q) whose scan shows total sc fat loss in the calves, major sc fat loss in the mid-thigh and marbled appear- ance of muscle tissue due to severe amounts of fat being stored within the muscle. waist-to-hip circumference (android pattern). BMI and right and left sides. The overall inter-observer correlation PBF for the total body and lower extremities for FPLD2 coefficients were, on average, 0.988 for the mid-calf and subject GL0096 was similar to young control GL2784 and 0.991 for the mid-thigh. significantly less than older control GL2990 (Table 1). Quantification of subcutaneous fat from MRI Qualitative differences on survey MRIs Quantification of the percent subcutaneous adipose tissue Qualitative coronal regional fat distribution profile differ- present in the mid-calf and mid-thigh regions showed that ences between affected and normal controls are shown in the control subjects had values ranging from 26-56%, Figure 1. The main visible differences included: 1) greater with mid-thigh values always greater than mid-calf values. subcutaneous fat depots, especially around the hips and The older control subject GL2990 (BMI 34.8) had percent thighs, for the two control subjects compared with the subcutaneous adipose tissue values that were ~1.3-fold FPLD3 and FPLD2 subjects; and 2) attenuation of subcu- greater than the younger, normal weight control subject taneous fat stores at a lower point on the thigh of the GL2784 P < 0.0001). The FPLD3 subject GL0658 had sig- FPLD3 subject compared to the FPLD2 subject. nificantly lower percent adipose tissue values for both the mid-calf and mid-thigh regions in comparison to both Intra- and inter-observer correlations for quantitative MRI control subjects (P < 0.0001 for both). The most signifi- analysis cant attenuation in subcutaneous adipose tissue was Intra-observer correlation was determined by comparing observed for the FPLD2 subject, where no subcutaneous two replicates of percent subcutaneous fat for both the connectedness map of fat was attainable for the minute mid-calf and mid-thigh derived from subjects GL2784, remnants of adipose tissue present in the perimeter of GL2990, GL0658 and GL0096. Each replicate involved mid-calf region, and thus quantification using the auto- analysis of 17 transaxial images for both the right and left mated Connected Threshold Grower tool was impossible. sides. Intra-observer correlation coefficients based on at The percent subcutaneous adipose tissue in the mid-thigh least 68 sections each were, on average, 0.996 for the mid- of FPLD2 subject GL0096 was also significantly lower calf and 0.998 for the mid-thigh. Inter-observer correla- than that observed for FPLD3 subject GL0658 (24.3 ± tion was determined by comparing percent subcutaneous 3.7% vs 34.4 ± 2.5%, P < 0.0001). fat for both the mid-calf and mid-thigh derived from sub- jects GL2784, GL2990, GL0658 and GL0096, as measured Mean and overall percent adipose tissue values are by two independent observers. Each determination reported (Table 1), representing averaged values of indi- involved analysis of 17 transaxial images for both the vidual slices composing each respective image stack and Page 5 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 sue in the thigh was 24.3 ± 3.7% and 34.4 ± 2.5% (P < 0.0001), for the FPLD2 and FPLD3 patients respectively. Current clinical assessment of adipose tissue distribution in common obesity and metabolic syndrome and subjects with FPLD2 and FPLD3 is still in its infancy. Also, BIA failed to capture differences in percent fat in lower extrem- ities in FPLD2 vs FPLD3 perhaps because so much fat was infiltrated into muscle in FPLD2. In contrast, MRI adipose connectedness maps and semi-automated subcutaneous adipose tissue quantification with very high resolution and reproducibility, captured traits that could be com- pared statistically, confirming the subtle clinical differ- ences [3,10]. This semi-automated method involved a Connected Threshold Grower tool which specified inclusion of only adipose tissue connected to the initial subcutaneous seed point. Based on this pilot study of FPLD patients, we observed very high intra- and inter-observer correlation values: r > 0.99 and >0.98, respectively. In addition to its reproducibility, the described method yields results quickly and accurately, with minimal user intervention. The method was limited by including only connected Quan Figure 3 tification of percent adipose tissue infiltrated adipose tissue. However, given the imprecise Quantification of percent adipose tissue. For each of definition of subcutaneous adipose tissue in extremities, the 17 transaxial slices in a given anatomical section, both the we elected to include the connected infiltrated adipose tis- total volume and the total subcutaneous (sc) and connected sue in our calculations, again since this would require no infiltrated (inf) fat volumes were selected using the Con- user judgment and/or intervention, thus reducing another nected Threshold Grower tool. Their corresponding vol- potential source of analytic variation. An additional limi- umes were determined using the Voxel Counter tool. The tation inherent in the ImageJ software, which does not percent adipose tissue was calculated for each slice by divid- affect reproducibility but affects image dynamic, is that of ing the total voxels determined for the sc + inf fat by the the 16-bit to 8-bit change to the image stacks prior to anal- total voxels for the slice. The percent adipose tissue was determined for each slice alone and also for the overall sec- ysis. This reduction in image dynamic, which reduces res- tion, combining the results from all 17 slices. olution, is a common setback in medical image processing where similar general-purpose software librar- ies are used. Future development of the software to utilize values of total volume subcutaneous adipose tissue original raw images would be advantageous in maintain- respectively. Each of these values is also an average of two ing image integrity and reflecting more accurate analysis replicate data sets from two independent analyses. The data acquired from quantification. correlation (r) between mean subcutaneous fat areas and overall fat volume was 0.99998. Evaluating FPLD patients theoretically allowed for assess- ment of the lower limits of resolution of the method; however, the method appeared insensitive for calf adipose Discussion Using a strategy to quantify subcutaneous fat in the lower measurements in FPLD2, since there was no subcutane- extremity that was based on connectivity analysis, we ous fat according to the definition specified in the quanti- found significant differences between subcutaneous adi- fication methodology. Future application of this pose tissue in the mid-calf and mid-thigh sections of FPLD quantification method may include quantification of patients compared to normal controls. We found signifi- both thigh and calf depots for "garden variety" obesity, cantly reduced lower extremity subcutaneous adipose tis- metabolic syndrome or diabetes. This approach might sue in a subject with FPLD2 than in a subject with FPLD3. also be applicable to quantify metabolically important Specifically, no subcutaneous adipose tissue could be substrata of fat [11]. quantified in the calf of the FPLD2 patient compared to 19.2 ± 1.7% subcutaneous adipose tissue in FPLD3 (P < We recognize that this study was limited due to the small 0.0001). Similarly, the percent subcutaneous adipose tis- subject numbers from whom subcutaneous adipose tissue Page 6 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 values were extracted. Acquisition of such values from a Competing interests larger number of patients with both FPLD subtypes would The author(s) declare that they have no competing inter- verify the likely results observed here. Furthermore, con- ests. trols were not ideally matched for age and BMI: while the FPLD2 patient had a similar BMI as the young control Authors' contributions individual, unmeasured and uncontrolled factors related SAA participated in the experimental design, data acquisi- to age might have further contributed to variation in sub- tion and analysis, interpretation of results, and manu- cutaneous adipose tissue. Expanding the sample size in script writing. RLP participated in the analysis of the MRI future studies would clearly be helpful in this regard. data and manuscript writing. JFR participated in data acquisition, analysis and interpretation of results. BAM The whole body scans suggested that this method can be was involved in the clinical assessment. RW performed adapted for other fat depots or bodily organs. However, the MRI scans. BKR participated in the experimental widespread application would depend on development of design, data analysis, and interpretation of results. RAH standards with respect to regions surveyed, anatomical participated in the experimental design, data analysis, landmarks, number of measurements, etc – similar to the interpretation of results and manuscript writing. All consensus standards agreed upon for carotid intima- authors approved the final manuscript. media thickness measurements using ultrasound. Also, intramuscular fat is distributed either in intra- or inter- Acknowledgements Supported by the Jacob J. Wolfe Distinguished Medical Research Chair, the myocellular depots; which could be more specifically Edith Schulich Vinet Canada Research Chair (Tier I) in Human Genetics, a evaluated using proton magnetic resonance spectroscopy Career Investigator award from the Heart and Stroke Foundation of (MRS) and/or fat selective MRI [12-14]. Such regional dis- Ontario, and operating grants from the Canadian Institutes for Health tribution could be an additional MRI analyte that could Research, the Heart and Stroke Foundation of Ontario (NA5320), the be considered together with other intermediate traits in Ontario Research and Development Challenge Fund (Project #0507) and subjects with FPLD or even common metabolic syn- by Genome Canada through the Ontario Genomics Institute. drome. Furthermore, it is possible to obtain carbon-13 nuclear magnetic resonance (NMR) spectra of human References muscle glycogen in vivo in diabetic patients [15], which 1. Garg A: Acquired and inherited lipodystrophies. N Engl J Med 2004, 350(12):1220-1234. has helped understand the pathogenesis of insulin resist- 2. Hegele RA: Phenomics, lipodystrophy, and the metabolic syn- ance, metabolic syndrome and type 2 diabetes. Quantifi- drome. Trends Cardiovasc Med 2004, 14(4):133-137. 3. Hegele RA: Lessons from human mutations in PPARgamma. cation of fat depots using MRI and appropriate image Int J Obes (Lond) 2005, 29 Suppl 1:S31-5. analysis software could provide complementary analytes 4. Agarwal AK, Garg A: A novel heterozygous mutation in perox- for research and perhaps eventually for the diagnosis and isome proliferator-activated receptor-gamma gene in a patient with familial partial lipodystrophy. J Clin Endocrinol monitoring of interventions. Metab 2002, 87(1):408-411. 5. Garg A, Peshock RM, Fleckenstein JL: Adipose tissue distribution pattern in patients with familial partial lipodystrophy (Dun- Conclusion nigan variety). J Clin Endocrinol Metab 1999, 84(1):170-174. In summary, we report the use of MRI and image analysis 6. Garg A, Vinaitheerthan M, Weatherall PT, Bowcock AM: Pheno- software employing Connected Threshold Grower and typic heterogeneity in patients with familial partial lipodys- trophy (dunnigan variety) related to the site of missense Voxel Counter tools to help quantify lower extremity sub- mutations in lamin a/c gene. J Clin Endocrinol Metab 2001, cutaneous fat depots in patients with two molecular forms 86(1):59-65. of partial lipodystrophy. We also showed that the meas- 7. Iacobellis G: Imaging of visceral adipose tissue: an emerging diagnostic tool and therapeutic target. Curr Drug Targets Cardi- urements showed high intra- and inter-observer correla- ovasc Haematol Disord 2005, 5(4):345-353. tion in a small sample. Finally, the measurements could 8. Liou TH, Chan WP, Pan LC, Lin PW, Chou P, Chen CH: Fully auto- mated large-scale assessment of visceral and subcutaneous be compared statistically and thus confirmed the clinical abdominal adipose tissue by magnetic resonance imaging. impression that FPLD2 and FPLD3 differ with respect to Int J Obes (Lond) 2006, 30(5):844-852. the extent of subcutaneous fat loss; specifically, subcuta- 9. ImageJ: Image processing and analysis in Java [http:// rsb.info.nih.gov/ij/] neous fat loss in the FPLD2 subject is greater than in the 10. Hegele RA, Cao H, Frankowski C, Mathews ST, Leff T: PPARG FPLD3 individual. Increasing the sample size of FPLD F388L, a transactivation-deficient mutant, in familial partial subjects in future studies will validate this interpretation. lipodystrophy. Diabetes 2002, 51(12):3586-3590. 11. Smith SR, Lovejoy JC, Greenway F, Ryan D, deJonge L, de la Bretonne These tools can be applied immediately and might be use- J, Volafova J, Bray GA: Contributions of total body fat, abdomi- ful in quantitative phenotype analysis of other forms of nal subcutaneous adipose tissue compartments, and visceral adipose tissue to the metabolic complications of obesity. lipodystrophy and in less extreme disorders of fat redistri- Metabolism 2001, 50(4):425-435. bution or repartitioning, such as "garden variety" obesity, 12. Boesch C, Slotboom J, Hoppeler H, Kreis R: In vivo determination insulin resistance, or type 2 diabetes. of intra-myocellular lipids in human muscle by means of localized 1H-MR-spectroscopy. Magn Reson Med 1997, 37(4):484-493. Page 7 of 8 (page number not for citation purposes) BMC Medical Imaging 2006, 6:11 http://www.biomedcentral.com/1471-2342/6/11 13. Brechtel K, Jacob S, Machann J, Hauer B, Nielsen M, Meissner HP, Matthaei S, Haering HU, Claussen CD, Schick F: Acquired general- ized lipoatrophy (AGL): highly selective MR lipid imaging and localized (1)H-MRS. J Magn Reson Imaging 2000, 12(2):306-310. 14. Schick F, Eismann B, Jung WI, Bongers H, Bunse M, Lutz O: Compar- ison of localized proton NMR signals of skeletal muscle and fat tissue in vivo: two lipid compartments in muscle tissue. Magn Reson Med 1993, 29(2):158-167. 15. Petersen KF, Shulman GI: Pathogenesis of skeletal muscle insu- lin resistance in type 2 diabetes mellitus. Am J Cardiol 2002, 90(5A):11G-18G. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2342/6/11/prepub Publish with Bio Med Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical researc h in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright BioMedcentral Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp Page 8 of 8 (page number not for citation purposes)

Journal

BMC Medical Imaging – Springer Journals

Published: Aug 31, 2006

There are no references for this article.