Multi-modal MR image brain tumor segmentation method based on deep learning and multi-guidance

Abstract

The invention discloses a multi-modal MR image brain tumor segmentation method based on deep learning and multi-guidance, belongs to the field of image processing, and solves three problems in a multi-modal MRI brain glioma segmentation process: (1) the problem of inaccurate segmentation caused by unclear brain glioma boundary; (2) some discrete wrong segmentation points appear in a segmentation result due to uneven brightness distribution of the multi-mode MRI; and (3) the problem of carrying out feature fusion on various kinds of guide information in a brain glioma MRI segmentation network.Feature fusion is carried out on an overall brain glioma segmentation result and a brain glioma edge prediction result through a proposed fusion mechanism, so that multi-mode MRI brain glioma segmentation under multi-feature map guidance and fusion is realized. According to the deep segmentation network, high-accuracy segmentation is realized with less parameter quantity, so that the method is convenient to embed into edge equipment to assist doctors in diagnosing and analyzing brain glioma.

Description

technical field [0001] The invention belongs to the field of image processing, in particular to a multi-modal MRI brain tumor segmentation method based on deep learning and multi-guidance. Background technique [0002] Glioma is a primary brain tumor, accounting for about 80% of malignant brain tumors. MRI assists doctors in diagnosing gliomas, and doctors can use MRI to locate gliomas, and then conduct quantitative analysis of gliomas to implement treatment plans such as radiotherapy and surgery. Generally, doctors diagnose and label gliomas through multimodal MRI, but it is quite time-consuming to manually label gliomas slice by slice. At the same time, different doctors have different levels of experience, which leads to large differences in the results marked for the same case, which is not conducive to the precise treatment of glioma. Therefore, the MR image labeling of glioma needs computer-assisted doctors to complete. [0003] At present, the image segmentation me...

AI Technical Summary

Problems solved by technology

[0005] The present invention aims to overcome the deficiencies of the existing technologies, solve the problem of low segmentation accuracy caused by uneven brightness distribution and blurred target boundaries in MRI, and provide a multi-modal MRI brain tumor segmentation method based on deep learning and multi-guidance. The invention fuses the overall glioma segmentation results and glioma edge prediction results through the proposed fusion mechanism, and realizes the multi-modal MRI glioma segmentation under the guidance and fusion of multi-feature maps

Images