A Practical Method for Muscles Extraction and Automatic Segmentation of Leg Magnetic Resonance Images

A method is introduced for leg muscles extraction and segmentation of MR images. For leg segmentation, an integrated algorithm is proposed for practical application. And the BPV (Block Pixel Variation) method is involved to adjust the inhomogeneity in the brightness distribution. Then the bones region is removed according to contours defined by the simplified deformable models. In addition, the spatial enclosing relationship between the tissue regions is abstracted as a feature for segmentation. Finally, the various noises remained in the ROI of the images are eliminated based on the specific models instead of filtering. The validity and accuracy are tested based on the MR images of legs, and the volumes of muscles are calculated accurately.