Dual Multi-Modal Feature Fusion Network for the Evaluation of Osteosarcoma

In this paper, we propose a dual multi-modal feature fusion osteosarcoma evaluation network to address the problem of efficient utilization of osteosarcoma medical images for tumor necrosis rate assessment. Firstly, we propose a dual evaluation network to utilize dual input information. Then, a structure called the multi-modal information fusion module is incorporated into the network to focus on key information and enhance the information flow between network layers. Based on this framework, we propose a loss function called osteosarcoma generalization loss to solve the problem of data imbalance in osteosarcoma images. We conducted extensive experiments to evaluate the proposed method. The results of comparative ablation experiments show that our newly proposed method improves both the evaluation effectiveness and performance compared to existing methods.