Automatic Segmentation of MRI Human Thigh Muscles
Ezak Fadzrin Ahmad Shaubari, Jamie Stewart McPhee, Hans Degens, Moi Hoon Yap · 2018
This paper outlines the procedure for automated segmentation of human thigh magnetic resonance images. Datasets that included sequential transverse-plane images (approximately 20 sequential images) were used to develop the framework and the images of mid-thigh were selected to establish the segmentation algorithm because these images contain essential parametric and geometric information that represents the subsequent images of the same dataset. Hence, if the mid-thigh can be decomposed accurately, it would significantly improve the prospects of decomposing all remaining transverse images within the dataset. The segmentation framework was designed based on the integration of statistical signal processing and medical imaging algorithms. The results showed good segmentation accuracy and fast processing time.