IPGRN: An Integrated Progressive Gated Refinement Network for Breast Tumor Analysis

The predictive analysis of disease indicators for breast tumor patients holds significant clinical relevance for physicians in diagnosis and treatment, and can benefit from the integration of consumer health technologies in smart healthcare systems. In clinical practice, vast amounts of monitoring data-such as mammography, breast ultrasound, and magnetic resonance imaging-are generated. Smart healthcare devices and remote monitoring systems facilitate the collection, sharing, and utilization of this data, supporting personalized healthcare. In this context, the selection and utilization of medical data features is a crucial task. While some models have demonstrated effectiveness in this area, they often suffer from limitations such as inadequate feature enhancement, insufficient cross-modal interaction, and poor generalization across diverse datasets. To address these issues, we propose an Integrated Progressive Gated Refinement Network (IPGRN). IPGRN is realized through the fusion of the Progressive Inheritance Shared (PIS) module with the Higher-Dimensional Hierarchical Gated Interactions (HD-HGI) module. This deep fusion network, employing a dual-path architecture, demonstrates outstanding performance in multimodal medical multi-index prediction, while exhibiting strong generalization capabilities. We present experimental results on five authentic datasets, empirically confirming the effectiveness of our proposed IPGRN. The source code of this paper can be found at https://github.com/HaoDavis/IPGRN.