Automatic intra-abdominal muscle and fat image segmentation method

Abstract

The invention provides an automatic intra-abdominal muscle and fat image segmentation method, which is used for simultaneously segmenting abdominal muscle tissues, visceral fat tissues and subcutaneous fat tissues. The method mainly comprises the following steps of: classifying pixels of an abdominal image to obtain a preliminary pixel value change range of fat tissues, muscle tissues and a background part; Secondly, performing further segmentation on the basis of the pixel value change range of the muscle tissue to obtain an outer contour part and an inner contour part of the muscle respectively, and finishing the segmentation of the muscle tissue based on the parts. Finally, the adipose tissue segmented in the first step is divided into two parts by the muscle tissue, and the visceral adipose tissue is located in the muscle tissue. The subcutaneous adipose tissue is located outside the muscle tissue. Compared with the prior art, the method has the advantages that all sliced musculartissues and adipose tissues are segmented at a time, the correct rate is high, and full-automatic segmentation is achieved.

Description

technical field [0001] The invention proposes an automatic intra-abdominal muscle and fat image segmentation method. Background technique [0002] The composition ratio of body fat to muscle tissue helps predict many diseases, especially those related to obesity. [0003] At present, the general software on the market is mostly two-dimensional for muscle and fat segmentation, and it is semi-automatic, which relies on the professional experience of doctors. At the same time, it needs to mark different tissues on each slice, which is too time-consuming and practicality needs to be improved. [0004] At present, automatic segmentation algorithms have been proposed, such as using FFD registration algorithm to segment muscle tissue, graph cut method, threshold method, etc. These methods also have their own defects. FFD registration is based on two-dimensional image segmentation, and each image needs to be recalculated, which takes a lot of time. The threshold and graph cut metho...

AI Technical Summary

Problems solved by technology

These methods also have their own defects. FFD registration is based on two-dimensional image segmentation, and each image needs to be recalculated, which takes a lot of time. The threshold and graph cut methods rely on the pixel prior value information of muscles and tissues, which increases errors. chance of

Images