DHGFN: dual hypergraph fusion network for brain tumor classification

Le résumé fourni par la source

To address the issues of feature redundancy and the neglect of channel features in Hypergraph Neural Network (HGNN)-based methods for brain tumor classification, this paper introduces a Dual Hypergraph Fusion Network (DHGFN). Specifically, a Spatial Attention-based Hypergraph Embedding Module (SAHEM) and a Channel Attention-based Hypergraph Embedding Module (CAHEM) are devised to effectively capture key regions of images and optimize the selection of channel features, respectively. Building upon these modules, a Dual Hypergraph Fusion Module (DHGFM) is further proposed, which achieves coordinated optimization of spatial and channel features through a multilevel feature fusion strategy. This strategy enhances the model's recognition accuracy and generalization capabilities. To validate the efficacy of the proposed method, systematic experimental evaluations were conducted on two public medical image datasets of brain tumors. The experimental results demonstrate that the proposed DHGFN outperforms existing mainstream approaches in classification performance.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.