Original Research Accurate Segmentation of Subcutaneous and Intermuscular Adipose Tissue From MR Images of the Thigh Vincenzo Positano, MSc, 1 * Tore Christiansen, MSc, 2 Maria Filomena Santarelli, PhD, 1 Steffen Ringgaard, MD, 3 Luigi Landini, PhD, 1,4 and Amalia Gastaldelli, PhD 1 Purpose: To describe and evaluate a computer-assisted method for assessing the quantity and distribution of adi- pose tissue in thigh by magnetic resonance imaging (MRI). Materials and Methods: Twenty obese subjects were im- aged on a Philips Achieva 1.5T scanner by a fast spin-echo (FSE) sequence. A total of 636 images were acquired and analyzed by custom-made software. Thigh subcutaneous adipose tissue (SAT) and bone were identified by fuzzy clus- tering segmentation and an active contour algorithm. Mus- cle and intermuscular adipose tissue (IMAT) were assessed by identifying the two peaks of the signal histogram with an expectation maximization algorithm. The whole analysis was performed in an unsupervised manner without the need of any user interaction. Results: The coefficient of variation (CV) was evaluated be- tween the unsupervised algorithm and manual analysis per- formed by an expert operator. The CV was low for all mea- surements (SAT 2%, muscle 1%, IMAT 5%). Limited manual correction of unsupervised segmentation results (less than 10% of contours modified) allowed us to further reduce the CV (SAT 0.5%, muscle 0.5%, IMAT 2%). Conclusion: The proposed approach allowed effective com- puter-assisted analysis of thigh MR images, dramatically reducing the user work compared to manual analysis. It allowed routine assessment of IMAT, a fat-depot linked with metabolic abnormalities, important in monitoring the effect of nutrition and exercise. Key Words: IMAT; SAT; fuzzy clustering; adipose tissue; fat; MRI J. Magn. Reson. Imaging 2009;29:677– 684. © 2009 Wiley-Liss, Inc. MAGNETIC RESONANCE IMAGING (MRI) is recognized as the most effective tool to perform assessment of fat distribution in the human body, joining patient safety with high image resolution (1). MRI is able to distin- guish between subcutaneous adipose tissue (SAT) and fat deposited around internal organs (VAT) in the ab- dominal region. This feature is important because ab- dominal fat distribution (ie, VAT/SAT ratio) is associ- ated with the development of all features of metabolic syndrome, the accompanying insulin resistance, and cardiovascular disease (2). More recently, intermuscu- lar adipose tissue (IMAT) has gained attention as a novel fat-depot linked with metabolic abnormalities (3). IMAT could be a potential contributor to glucose dis- posal and muscle function (4) and could represent an important index in monitoring the effect of nutrition and exercise (5,6). The thigh is the preferred anatomical location for IMAT assessment because the mid-thigh volume is the best single predictor for whole-body IMAT (7). Although analysis of a single MRI slice may give the advantage of simple acquisition and postprocessing (8), accurate determination of fat distribution requires mul- tislice imaging and analysis. Manual analysis of 3D datasets is time-consuming and operator-dependent, and thus automatic or semiautomatic computer as- sisted methods are highly desirable. Several automatic and semiautomatic image analysis algorithms were de- veloped for abdominal fat distribution assessment (9 – 12), and several studies provided comparison between available softwares (13–15). General-purpose semiau- tomated softwares as NIH ImageJ (14), Slice-O-Matic (15), and Analyze (14) are widely validated and allow delineation of fat deposits in different body locations, but may require a long processing time. Automated methods show a good overall agreement with validated general-purpose softwares (14) and seem to have some distinct advantages over less automated methods, as there is less variability between evaluators and reduced analysis time (15). Although the importance of IMAT evaluation is also recognized, development of dedicated computer-as- sisted tools for assessment of fat distribution in the extremities is still in an early stage; hence, image anal- ysis is usually performed manually. Therefore, IMAT evaluation in clinical practice may require a long image processing time and may be affected by observer-in- duced variability. 1 MRI Laboratory, “G. Monasterio” Foundation and Institute of Clinical Physiology, Pisa, Italy. 2 Department of Medicine and Endocrinology C, Aarhus University Hos- pital, Aarhus, Denmark. 3 MR Research Center, Aarhus University Hospital, Aarhus, Denmark. 4 Department of Information Engineering, University of Pisa, Pisa, Italy. *Address reprint request to: V.P., Gabriele Monasterio Foundation, CNR, Via Moruzzi, 1, 56124, Pisa, Italy. E-mail: positano@ifc.cnr.it Received August 20, 2008; Accepted December 2, 2008. DOI 10.1002/jmri.21699 Published online in Wiley InterScience (www.interscience.wiley.com). JOURNAL OF MAGNETIC RESONANCE IMAGING 29:677– 684 (2009) © 2009 Wiley-Liss, Inc. 677