Congratulations are in order for the paper titled “LVM-Med: Learning large-scale self-supervised vision models for medical imaging via second-order graph matching” (Authors: Duy M. H. Nguyen, Hoang Nguyen, Nghiem T. Diep, Tan N. Pham, Tri Cao, Binh T. Nguyen, Paul Swoboda, Nhat Ho, Shadi Albarqouni, Pengtao Xie, Daniel Sonntag, and Mathias Niepert), which has been published in NeurIPS 2023.

The authors include both students and recent graduates engaged in this research:

1. Nguyễn Hồ Minh Duy, a former student of the Faculty of Mathematics — Computer Science and currently a Master’s student in Machine Learning at the Max Planck Institute, Germany (Co-first author);
2. Nguyễn Minh Hoàng, a fourth-year student at the Faculty of Mathematics — Computer Science (Co-first author);
3. Cao Thiên Trí, an alumnus of the Advanced Programme in Computer Science from the Faculty of Information Technology;
4. Phạm Ngọc Tân, a Master’s student in Artificial Intelligence;
5. Diệp Tường Nghiêm, a second-year student in the Advanced Programme in Computer Science at the Faculty of Information Technology.

The paper addresses the complexities involved in designing pre-trained models for medical images, which can be particularly challenging due to domain shifts. For instance, CT, MRI, and X-ray images possess differing characteristics that complicate the development of a unified model. The proposed solution, LVM-Med, is a contrastive learning algorithm grounded in high-order graph matching. This algorithm demonstrated an improvement of approximately 6 to 7% in accuracy compared to existing methods, proving applicable to challenging problems such as brain tumour classification and diabetic retinopathy grading.

• Link to the paper: https://arxiv.org/abs/2306.11925