Medical Engineering & Physics 32 (2010) 926–933 Contents lists available at ScienceDirect Medical Engineering & Physics journal homepage: www.elsevier.com/locate/medengphy Technical note Technique for the computation of lower leg muscle bulk from magnetic resonance images Barry J. Broderick a,b,* , Sylvain Dessus c , Pierce A. Grace d , Gearóid ÓLaighin a,b a Electrical & Electronic Engineering, National University of Ireland Galway, University Road, Galway, Ireland b National Centre for Biomedical Engineering Science, National University of Ireland Galway, University Road, Galway, Ireland c ENSEEIHT, Institut National Polytechnique de Toulouse, France d Department of Vascular Surgery, Mid-Western Regional Hospital, Limerick, Ireland article info Article history: Received 3 November 2009 Received in revised form 22 June 2010 Accepted 24 June 2010 Keywords: Calf muscle bulk Unsupervised segmentation Image processing Magnetic resonance imaging abstract An unsupervised technique to estimate the relative size of a patient’s lower leg musculature in vivo using magnetic resonance imaging (MRI) in the context of venous insufficiency is presented. This post- acquisition technique was designed to segment calf muscle bulk, which could be used to make inter- or intra-patient comparisons of calf muscle size in the context of unilateral leg ulcers and venous return. Pre-processing stages included partial volume reduction, intensity inhomogeneity correction and con- trast equalization. The algorithm created a binary mask of voxels that fell within a computed threshold designated as representing muscle based on a 3-class fuzzy clustering approach. The segmentation was improved using a set of morphological operations to remove adipose tissue, spongy bone and cortical bone. The technique was evaluated for accuracy against a manual segmented ground truth. Results showed that the automatic technique performed sufficiently well in terms of accuracy and efficacy. The automatic method did not suffer from intra-observer variability. © 2010 IPEM. Published by Elsevier Ltd. All rights reserved. 1. Introduction The recognition and delineation of objects of interest in medical images remains a challenging problem. Magnetic resonance image (MRI) segmentation has been extensively used in the past to study brain pathology [1–5] but has also been used for other tasks such as adipose tissue volume estimation [6–8], muscle tissue volume estimation [9,10], tumor volume determination [11,12] and detec- tion [13]. Despite the many different segmentation algorithms described in the literature, no one segmentation algorithm can yield acceptable results in every application domain [6,14]. Therefore, there remains a need for computer-assisted segmentation algo- rithms in various research areas. Calf muscle pump function is of particular interest to our research group. It is likely that calf muscle pump hemodynamic performance is directly related to the strength of the calf muscle contraction; the stronger the contraction, the greater the volume of blood emptied from the venous sinuses as a result. It is suspected that a reduction in calf muscle bulk through atrophy could result in a decrease in venous return, eventually lead- ing to chronic venous insufficiency. Muscle atrophy in sedentary or * Corresponding author at: Electrical & Electronic Engineering, National Univer- sity of Ireland Galway, University Road, Galway, Ireland. Tel.: +353 91 493126. E-mail address: b.broderick2@nuigalway.ie (B.J. Broderick). immobile patients can lead to a compromised calf muscle pump [15]. Thus, the application of a specific computerized calf muscle tissue segmentation algorithm could play an essential role in the study of patients with chronic venous insufficiency. This paper outlines a segmentation process for determining calf muscle bulk quickly and accurately from two-dimensional MRI slices of the lower leg. MRI is an invaluable medical resource for diagnostic purposes [16–19]. Various tissue types can be differenti- ated from each other using MRI by the variation in signal intensity. Fat-based tissues, such as adipose and bone marrow, have much higher signal intensities than water-based tissues, such as muscle or cartilage. The contrast in signal intensities defines the bound- aries between tissues [20]. Defining a threshold intensity value is the simplest method for evaluating tissue distributions. However, Elbers et al. [21] point out that absolute signal intensities vary from image to image and from individual to individual, and as a result, intensity threshold levels have to be set for each image. Manual or semi-automated segmentation of MR images is very time-consuming, is affected by user variability [7] and typically requires the input from experts with knowledge of anatomy. Earlier work into muscle volume measurements described by Mitsiopou- los et al. [10] and Elliott et al. [9] demonstrated that MRI was indeed a useful tool for measuring skeletal muscle. However, the described techniques required a substantial amount of user assistance which may incorporate a significant operator bias. 1350-4533/$ – see front matter © 2010 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2010.06.008