CVMAIL at MICCAI 2025: Keynote and Research Presentation
The Computer Vision and Medical AI Lab (CVMAIL), VinUniversity, is pleased to announce our active participation in MICCAI 2025, to be held in Daejeon, South Korea.
The International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI) is one of the world’s leading scientific conferences in medical imaging, computer-assisted intervention, and healthcare artificial intelligence. Established in 1998, MICCAI brings together researchers, clinicians, engineers, and industry leaders from around the globe to present cutting-edge advances in image analysis, machine learning, surgical technologies, and translational AI for medicine. The conference is highly competitive and internationally recognized for its rigorous peer-review process, serving as a premier platform for advancing innovation, collaboration, and clinical impact in medical image computing.
As part of the MICCAI 2025 Open Data Initiative, Dr. Pham Huy Hieu, Principal Investigator of CVMAIL and representative of the VinUni-Illinois Smart Health Center (VISHC), has been invited to deliver a Keynote Speaker. His talk is titled “Collection, Standardization, and Sharing of Large-Scale Annotated Datasets.”
In this keynote, he emphasized that diverse and representative medical datasets are essential not only for developing robust and generalizable AI models, but also for reducing healthcare inequalities across populations and regions. Together with leading experts – Associate Professor Mohammad Yaqub (MBZUAI; Board Member, MICCAI Society; General Chair, MICCAI 2026) and Dr. Spyridon Bakas (Indiana University School of Medicine; Board Member, MICCAI Society; Executive Editor at MELBA), Dr. Hieu engaged the MICCAI community in an in-depth discussion on fostering cross-institutional data sharing, particularly involving private sector stakeholders; the emerging role of synthetic data and evolving concepts of data ownership; and responsible hospital data governance that both protects patient privacy and enables impactful research. The session highlighted the urgent need for thoughtful, forward-looking policies that promote open science and innovation while ensuring ethical standards and data protection in the rapidly evolving landscape of healthcare AI.
In addition to the keynote, CVMAIL Research Assistant Ha Hieu Pham will present our latest research entitled: “Learning Disentangled Stain and Structural Representations for Semi-Supervised Histopathology Segmentation.”
This work proposes a novel representation learning framework that separates stain-related variations from structural tissue information, improving robustness and segmentation performance in semi-supervised histopathology settings. The study addresses key challenges in domain variability and limited annotations, contributing to more reliable AI-assisted pathology analysis.
CVMAIL’s participation at MICCAI 2025 reflects our continued commitment to advancing safe, generalizable, and data-centric AI for medical imaging. We look forward to engaging with the international research community in Daejeon and exploring new opportunities for collaboration.