Anatomical structure priori guided brain region-of-interest quick segmentation method and system

Abstract

The invention discloses an anatomical structure priori guided brain region of interest quick segmentation method and system. The method comprises the following steps of 1, learning brain nuclear magnetic resonance image features through a segmentation sub-network; 2, learning brain anatomical structure information through an anatomical structure attention sub-network; 3, fusing the image featureslearned by the segmentation sub-network and the anatomical structure information learned by the anatomical structure attention sub-network through an anatomical structure door; and step 4, segmentingthe brain nuclear magnetic resonance image by using the output of the segmentation sub-network. According to the method, the convolutional network is used to learn features, the anatomical structure prior information is used to guide the network to perform brain image segmentation, and the method provided by the invention can segment the brain nuclear magnetic resonance image more quickly and accurately.

Description

technical field [0001] The invention relates to a method and system for rapid segmentation of brain regions of interest guided by anatomical structure prior, and belongs to the technical field of medical image analysis. Background technique [0002] In recent years, deep learning methods have achieved great success in medical image segmentation and computer-aided diagnosis. In particular, some methods based on end-to-end network structures are often used to automatically segment images. An end-to-end network usually consists of two parts: (1) encoding part and (2) decoding part. Specifically, the encoding part is used to extract the high-level texture features of the input image, and the decoding part converts the high-level texture features of the image into a dense label set through an upsampling method, which is used to predict the label of the image to be segmented. In early end-to-end network structures, fully connected layers are used to convert high-dimensional feat...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method and system for rapid segmentation of brain regions of interest guided by anatomical structure priors, to solve the problem of low precision of deep network segmentation, slow speed of multi-atlas segmentation methods, and difficult manual definition of features in the prior art. question

Images