Nature Communicationsに当研究室の論文が採録

■書誌情報
Tianyu Liu, Weihao Xuan, Hao Wu, Peter Humphrey, Marcello DiStasio, Mohamed Kahila, Alfonso Garcia Tan, Heli Qi, Rui Yang, Simeng Han, Tinglin Huang, Fang Wu, Chen Liu, Qingyu Chen, Nan Liu, Irene Li, Hua Xu, Hongyu Zhao: Building MultiModal Pathology Experts with Reasoning AI Copilots, Nature Communications, 2026
■概要
Artificial intelligence is advancing computational pathology toward multimodal diagnosis, analysis, and interpretation. However, pathology-specific vision-language models often lack rigorous diagnostic reasoning and the flexibility to address diverse tasks, limiting their use as clinical copilots. We present TeamPath, an AI system that combines reinforcement learning with router-based task selection and trains on large-scale multimodal histopathology datasets. TeamPath supports expert-level disease diagnosis, patch-level information summarization, and cross-modal generation that integrates transcriptomic information for clinical applications. In collaboration with pathologists at Yale School of Medicine, we show that TeamPath improves workflow efficiency by identifying and correcting diagnostic conclusions and reasoning paths. Human evaluation further supports the quality and reliability of its reasoning. By dynamically selecting appropriate strategies for different needs, TeamPath provides an adaptable and dependable framework for communication across modalities and between AI systems and clinical experts, enabling more efficient, interpretable, and collaborative computational pathology workflows across diverse research and clinical settings.