DEVELOPMENT AND APPLICATION OF SEGMENTATION AND ANALYSIS METHODS FOR THE MORPHOLOGICAL AND FUNCTIONAL CHARACTERIZATION OF THE LOWER LIMB FROM MRI IMAGES

This work is aimed to the development and application of a segmentation and analysis methods for the morphological and functional characterization of the lower limb from MRI images. Regarding the morphological characterization, an automatic algorithm was developed for the segmentation of Skeletal Muscle (SM), Intermuscular Adipose Tissue (IMAT), and Subcutaneus Adipose Tissue (SAT) compartments from cross-sectional T1-W MRI images, in order to assess thigh regional tissue composition in both young and elderly subjects with different degrees of body adiposity, including obese individuals. A fuzzy c-mean algorithm was employed to perform a classification of the different tissues: SM, Adipose Tissue (AT) and bone. Muscle fascia, which is the anatomical structure that separates SAT and IMAT compartments, was segmented using a procedure based on snake active contour model. We validated the segmentation framework on 15 datasets from 5 young normal weight, 5 older normal weight and 5 older obese females using manual segmentations delineated by an expert operator as gold standard. Segmentation errors were assessed for each structure resulting in mean relative area difference of 1.8%, 2.5% and 2.7% for SM, bone and AT, respectively, and a mean sensitivity for each compartment above 96% in each subject typology. Muscle fascia identification performance resulted in a mean distance between manual and automatic contours of 0.81mm and a mean percentage of contour pixels with distance smaller than 2 pixels of 86.2%. Moreover, manual and automatic IMAT and SAT cross-sectional areas in all subject typologies were found significantly correlated (p < 0.001). These results indicate that the proposed automatic segmentation algorithm, adequately performing thigh tissue discrimination, could be an helpful tool in studies of thigh regional composition. To our knowledge, this is the first published approach which identifies muscle fascia in its anatomical position obtaining promising results from a low level based segmentation perspective. Regarding the functional characterization of the lower limb, the properties of a specific region of the SM: the quadriceps femoris was considered representative of the entire compartment. In this region, size and fat content of 6 normal weight and 5 obese well functioning older females were determined at mid-thigh of the dominant leg, by analysing MRI Dixon images. These data as well as peak knee-extension torque, muscle fascicle length and pennation angle were compared in order to assess functional differences between the two groups. The study showed that muscle tissue composition and pennation angle are important determinants of muscle torque per unit muscle section in well-functioning older women. Moreover, as a result of the greater gravitational load, active obese elderly women have more muscle mass but with a higher fat content compared with normal-weight counterparts. The amount and distribution of SM and AT in different body regions, have a relevant clinical impact. In particular, the assessment of changes in both SM and AT amount and distribution are significant as they correlate with processes related to aging. Namely, a loss of SM mass with an increasing of IMAT have been identified as a negative correlate of SM quality and strength in lower limb, leading to functional impairment of different severity. Moreover, the simultaneous presence of such conditions with an abnormal high accumulation of AT in the lower limb has been defined as sarcopenic obesity and correlated with accelerated functional decline with high risk of diseases and mortality. This work represents a step forward not only to the accomplishment of accurate patient-specific thigh tissues segmentation and quantification, but also to the understanding of processes related to aging with the concomitant presence of obesity.