Multi-modal feature fusion MRI brain tumor image segmentation method based on attention mechanism

Abstract

The invention discloses a multi-modal feature fusion MRI brain tumor image segmentation method based on an attention mechanism, and relates to the field of deep learning, and the method comprises the steps: firstly carrying out the data preprocessing and data augmentation of a data set, then constructing a network model, and enabling the network model to comprise a backbone network, a mixed context sensing module and a global attention fusion module, an image enters a network model and is coded through a backbone network, then global and local information is sensed through a hybrid context sensing module, and finally multi-modal features are fused through an attention fusion module and the image is output. And after passing through the trained network model, inputting a to-be-segmented two-dimensional magnetic resonance brain tumor image into the trained model, and outputting an image segmentation result. According to the method, the effective network model for automatically segmenting the MRI brain tumor image can be trained, the multi-modal features are fused, the segmentation precision is improved, and the method has relatively high application value and application prospect of clinical treatment.

Description

technical field [0001] The invention belongs to the technical field of deep learning and is applied to medical image segmentation, in particular to an MRI brain tumor image segmentation method based on multimodal feature fusion based on an attention mechanism. Background technique [0002] Brain tumor segmentation is critical for the diagnosis and prognosis of glioma patients. Segmenting brain tumors from magnetic resonance images is an essential procedure in brain tumor treatment, enabling clinicians to identify tumor location, extent, and type. This not only aids in initial diagnosis, but also in management and monitoring of treatment progress. Given the importance of this task, precise characterization of tumors and their subregions is often done manually by experienced neuroradiologists. This is a tedious and time-consuming process that requires a lot of time and expertise, especially to segment images of patients whose tumors are large, images are multimodal, and tumo...

AI Technical Summary

Problems solved by technology

Multiple modal image information can effectively complement each other, which can effectively improve the accuracy of segmentation, but it also increases the difficulty of the segmentation problem to a certain extent. The input multi-modal image information increases the necessary information for segmentation, but at the same time increases a large amount of Unnecessary information, thus making the segmentation problem more difficult

Images