Brain tumor image generation and segmentation method based on deep neural network

Abstract

The invention relates to a tumor image segmentation method based on image generation, and belongs to the technical field of medical image analysis. Aiming at a scene in which modal loss exists in a multi-modal medical image, a two-stage multi-task framework is adopted to realize the generation of a missing modal and complete the segmentation of a target area, and the method is realized by the following technical scheme: firstly, a multi-task generator is used to simultaneously obtain a pseudo target modal image and a preliminary segmentation result; the quality of the generated image is improved through a global discriminator and a local discriminator, and meanwhile, a perception loss function is provided to reduce the semantic difference between the generated target domain image and the real target domain image. Secondly, the invention provides a multi-task fine segmentation network, on the basis of the generated target domain image and the preliminary segmentation result, errors in the fine segmentation result and the preliminary segmentation result are predicted at the same time, and the consistency constraint between the two predictions is introduced to improve the segmentation performance. Compared with direct segmentation from an original partial modal image, the method has the advantages that the segmentation precision is greatly improved, and the method is superior to an existing image generation and segmentation method.

Description

technical field [0001] The invention relates to a method for generating and segmenting a medical image, in particular to generating a missing mode of a tumor and segmenting a tumor region, and belongs to the technical field of medical image analysis. Background technique [0002] Medical images are of great value in the diagnosis of tumors. Extracting the boundaries of tumor regions from medical images is a crucial step in the accurate measurement of the three-dimensional volume and shape of tumors, as well as in the planning of surgery and radiotherapy. Brain and other central nervous system (CNS) tumors are among the most common types of cancer, with an estimated annual incidence of 29.9 per million adults, of which approximately one-third are malignant. The development of medical imaging technology provides a reliable way for the diagnosis of brain tumors, such as glioma and acoustic neuroma. The automatic delineation and analysis of brain tumors from images will help red...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to overcome the deficiencies in the existing brain tumor image generation and segmentation technology, and to solve the problem of brain tumor segmentation in the case of partial modality loss, a novel two-stage multi-task consistency framework is proposed to realize from Generation of source image modality to target modality image, and further obtain tumor segmentation results

Images