Research
The following summarizes MedAI’s research activities and representative outcomes in several research domains.
Evidential Feature Representation of Heterogeneous Information
Focuses on confidence modeling, evidence combination, and knowledge integration for heterogeneous multimodal data, with representative publications in TMI, NeurIPS, ICLR, AAAI, BIBM, etc.
Evidence-driven Multidimensional Uncertainty Disentanglement
Focuses on conflict, ambiguity, and unknownness quantification through evidential learning frameworks, with representative publications in TMI, MIA, NeurIPS, AAAI, MICCAI, etc.
Statistically Calibrated Decision Risk Optimization
Focuses on conformal prediction and multiple hypothesis testing for risk-aware decision-making, with representative publications in TCSVT, MIA, CVPR, ACM MM, AAAI, etc.
Full list of publications in Google Scholar
Clinical AI Systems (Selected Cases)
Focuses on the development of clinical AI systems, including a pancreatic tumor-assisted diagnosis system and an endodontic diagnosis support system (TJ-ToothAI), enabling real-world deployment of trustworthy intelligent medical solutions.
Pancreatic Tumor-Assisted Diagnosis System
Res. Chao Ma, Changhai Hospital (Co.)
Prof. Siming Zheng, The First Affiliated Hospital of Ningbo University (Co.)
Endodontic Diagnosis Support System (TJ-ToothAI)
Prof. Qi Zhang, The Affiliated Stomatology Hospital of Tongji University (Co.)
