Full-type aortic dissection true and false cavity image segmentation method and system

Abstract

The invention relates to a full-type aortic dissection true and false cavity image segmentation method and system. The method comprises the steps: segmenting a to-be-segmented true and false cavity image through a segmentation network model based on an attention mechanism, mapping an obtained mask image into the to-be-segmented true and false cavity image, and obtaining a segmentation result; in the process of training the segmentation network model, marking an aorta region, a true cavity and a false cavity, extracting aorta dissection spatial correlation of true and false cavity data by using 3D convolution, and performing up-and-down sampling by using a mixed attention module as a basic unit to apply a weight to the true and false cavity data, and completing prediction of each pixel value category is completed through exponential function normalization. The problem of non-full automation in the prior art is solved on the premise that the segmentation precision is ensured, and the aortic dissection true and false cavity segmentation neural network model based on the attention mechanism is put forward, so that the model is more sensitive to the distinguishing of true and false cavities, and the precision of the model is improved.

Description

technical field [0001] The invention relates to the technical field of image segmentation, in particular to a method and system for segmenting true and false lumen images of all types of aortic dissection. Background technique [0002] Existing end-to-end image segmentation methods for true and false lumen of aortic dissection take 3D volume data composed of a series of CT images as input, and realize the segmentation of true and false lumen of aortic dissection through various semantic segmentation networks. The geometric structure of the aorta is a circular lumen, but due to its curved state in the human body, the CT image sequence is not completely regular circle, which brings difficulties to the segmentation, so the non-end-to-end aortic dissection segmentation method is increased. In order to extract the centerline of the aorta, a three-dimensional image of the straightened aorta was obtained through multi-plane reconstruction along the centerline. This type of researc...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method and system for image segmentation of true and false lumens of all types of aortic dissection to solve the current problem of low segmentation accuracy or non-automatic segmentation process

Images