Medical Image Graph Cut Segmentation Method Based on Statistical Shape Model

Abstract

The invention discloses a medical image Graph Cut segmentation method based on a statistical shape model, which mainly solves the problem in the prior art that low-contrast organs are difficult to segment well in medical images. The implementation steps are: (1) establish a statistical shape model of low-contrast organs and collect grayscale information; (2) pre-segment low-contrast organs; (3) initialize Graph; (4) segment low-contrast organs. The medical image Graph Cut segmentation method based on the statistical shape model of the present invention uses the Graph Cut algorithm for rapid segmentation, and adds the prior knowledge of the shape of the organs to reduce the possibility of over-segmentation and under-segmentation, and utilizes animal organs and animal The relative relationship between in vitro contours determines the initial position of low-contrast organs and improves segmentation efficiency. It is a fast and effective organ segmentation method.

Description

technical field [0001] The invention belongs to the field of medical imaging and relates to a statistical contour model and graph cut segmentation. Background technique [0002] As a technology that can provide human body structure information in a non-invasive manner, medical image segmentation is the first step in the analysis and visualization of medical image data. Pathology location, tracking disease progression, and determining radiation therapy dose or surgery size. Therefore, medical image segmentation is a key step in the computer-assisted therapy system, and the accuracy of segmentation directly affects the effect of auxiliary therapy. [0003] However, accurate segmentation of medical images faces many challenges. First, many anatomical structures are non-uniform under the spatial overlap of pixel or voxel gray levels. Secondly, there are low-contrast parts in medical images. For example, the edges of kidneys and hearts are difficult to distinguish in CT images...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to propose a segmentation method based on a statistical shape model and a Graph Cut method for the problems of inaccurate segmentation and low efficiency in the above-mentioned medical image organ segmentation.

Aiming at the problem that the Graph Cut segmentation method cannot be accurately segmented when the difference between the target organ and the background is small, the present invention adds a shape prior by introducing a statistical shape model, and combines the statistical shape information of the organ to be segmented to assist segmentation

Images