主讲嘉宾：英国谢菲尔德大学，陈晨

时       间：2025年12月15日，16:30-17:30

地       点：深圳大学医学部生物医学工程学院517大会议室

主  持  人：杨鑫

嘉宾简介:

       Chen (Cherise) Chen is currently an Assistant Professor in Computer Vision at the School of Computer Science, University of Sheffield, UK. She is also an ELLIS Scholar in Robust Machine Learning at the European Laboratory for Learning and Intelligent Systems. Her research focuses on the intersection of AI and healthcare, particularly in developing data-efficient, robust and explainable AI for clinical applications.

       Previously, she was a postdoc at Imperial College London (ICL) and University of Oxford. She obtained her Ph.D. from ICL in 2022. Chen also had valuable industrial experience as a research scientist at Infervision Inc., Beijing and HeartFlow, UK. So far, she has published more than 40 papers in leading conferences and high-impact journals such as MICCAI, ICCV, ECCV, as well as European Heart Journal, IEEE TMI and Medical Image Analysis, accumulating more than 4,000 Google Scholar citations with an h-index of 23. She is also the program chair for MIDL 2025; area chair for MICCAI 2024-2025 and workshop chair for BMVC 2025, and has organized several MICCAI workshops and challenges including, MICCAI ADSMI 2024, DALI 2023 and the CMRxMotion Challenge.

报告摘要:

       Recent years have brought rapid progress in AI through larger datasets and stronger computation, leading to major advances in multimodal and generative models. However, in the medical domain, we often deal with limited data and annotations with class imbalance, limited compute, missing modalities, and strict deployment requirements. In this talk, I will present our recent work on multimodal AI for healthcare that looks at these challenges as design opportunities. I will show how generative models can help build more robust cardiac scar segmentation models, how multimodal learning can strengthen single-modality analysis when data are scarce, and how to design models that continue to work when some inputs are missing.