Fine-grained Prototype Network for MRI Sequence Classification

The proposed new MRI sequence classification model, SequencesNet, addresses the problem of subtle inter-class differences and significant intraclass variations existing in medical images. The modular design of SequencesNet can be extended to other medical imaging tasks, including but not limited to multimodal image fusion, lesion detection, and disease staging. Future work can be done to decrease the computational complexity and increase the generalization of the model.