Automatic Segmentation Method of Abnormal Areas in Lung CT Images

Abstract

The method for automatically segmenting the abnormal region of the lung CT image provided by the present invention includes: searching for the minimum value of the neighborhood gradient of each point from the CT image of the lung parenchyma through a slide-down algorithm, and obtaining the region of the abnormal initial growth point of the lung according to the minimum value of the neighborhood gradient , determine the initial growth seed point according to the area of ​​the abnormal initial growth point of the lung; obtain the pixel grayscale constraint and the growth distance constraint according to the initial growth seed point, and determine the lung abnormality by the region growing method from the pixel grayscale constraint and the growth distance constraint area; according to the lung abnormal area, the boundary of each layer of lung abnormal area is obtained, and then the boundary pixel points of the adjacent layer in the lung abnormal area in the preset direction are obtained, and the boundary pixel points of the adjacent layer and the lung The center point of the abnormal region of the lungs is used to obtain the average distance difference, and the boundary pixel points whose distance difference exceeds the average distance difference among all the boundary pixels are smoothed to obtain the segmented image of the abnormal lung region.

Description

technical field [0001] The invention relates to image segmentation technology, in particular to an automatic segmentation method for abnormal areas of lung CT images. Background technique [0002] The automatic segmentation of abnormal lung CT images is an important issue in the field of computer-aided diagnosis. It can automatically obtain abnormal lung tissue and shorten the waiting time for clinical diagnosis. It has important application value in the field of computer-aided diagnosis. [0003] Abnormal lung CT images can be classified into isolated type, adhesion with blood vessels, and adhesion with pleura according to the abnormal location; according to the abnormal type, it can be divided into ground glass opacities, pulmonary nodules, and lung tumors. Among them, the ground-glass opacity type is the most difficult to separate among the lung abnormalities. The existing segmentation algorithms for abnormal lung tissue based on region growth all need to manually obtain...

AI Technical Summary

Problems solved by technology

Existing segmentation algorithms for abnormal lung tissue based on region growth require manual acquisition of initial growth seed points, resulting in inaccuracy

Images