A Brain Tumor Segmentation Method Based on Multi-Hierarchical Relational Learning Network

Abstract

The invention proposes an advanced multi-level structure relation learning network for segmenting brain tumor data. In each sub-network, an environmental information mining module is introduced between the encoder and the decoder, and dual self-attention mechanism and spatial interaction learning are used to mine environmental information within a single domain and between different domains, respectively.

Description

technical field [0001] The invention relates to tumor segmentation technology, in particular to a brain tumor segmentation method based on a multi-level structure relationship learning network. Background technique [0002] Segmentation of brain tumors, segmenting different types of tumor regions in multimodal 3D magnetic resonance images. Brain tumor segmentation based on MRI data is an important academic and industrial topic and has been an active area of ​​research over the past decade. Efficient and rapid brain tumor segmentation facilitates neurological status monitoring, tumor development assessment, and encephalopathy diagnosis. [0003] In recent years, deep learning-based cascaded multi-layer networks and multi-scale analysis have made great progress in medical image segmentation. However, in brain tumor segmentation based on MRI data, how to accurately classify each pixel remains a challenge. MRI data images have low contrast and different types of tumors have a...

AI Technical Summary

Problems solved by technology

[0005] For the second problem, the independent analysis and linear post-fusion of different domains make the segmentation model easy to fall into the local optimal problem

Images