Human vertebral body slice muscle fat segmentation method, system, electronic device and medium

Abstract

The application provides a human vertebral body layer muscle fat segmentation method, system, electronic equipment and medium; the method comprises the following steps: acquiring a three-dimensional CT medical image of a human vertebral body; resampling the three-dimensional CT medical image to obtain a three-dimensional resampled image; generating a semantic segmentation three-dimensional mask covering all regions of the three-dimensional resampled image based on the three-dimensional resampled image; along the Z-axis of the semantic segmentation three-dimensional mask, the proportion of vertebral body category pixels of all vertebral body categories contained in the semantic segmentation three-dimensional mask is counted slice by slice to generate a one-dimensional dominant vertebral body label sequence based on the proportion of vertebral body category pixels; based on the one-dimensional dominant vertebral body label sequence, a morphological post-processing is adopted to obtain a vertebral body layer positioning mask; and the muscle fat in the human vertebral body layer is segmented according to the vertebral body layer positioning mask; the application solves the problem that the traditional method relies on end-to-end layer labeling and has high calculation complexity, and still maintains sub-millimeter level positioning accuracy in a low computing power scene, thereby being beneficial to muscle fat segmentation.

Images