Mamba2MIL: State Space Duality Based Multiple Instance Learning for Computational Pathology
Published in arXiv preprint arXiv:2408.15032, 2024
Mamba2MIL studies multiple instance learning for computational pathology with an emphasis on modeling long patch sequences from whole-slide images. The work combines state space duality with weighted feature selection and sequence transformation to build more expressive slide-level representations.
This project reflects my research interest in multimodal and representation-learning methods for medical imaging, especially computational pathology.
Related links:
Recommended citation: Yuqi Zhang, Xiaoqian Zhang, Jiakai Wang, Yuancheng Yang, Taiying Peng, and Chao Tong. (2024). "Mamba2MIL: State Space Duality Based Multiple Instance Learning for Computational Pathology." arXiv preprint arXiv:2408.15032.
Download Paper