VinUni-Illinois Smart Health Center Members Win First Place at CXR-LT 2026 Challenge

The students from the VinUni-Illinois Smart Health Center (VISHC) have secured First Place at the prestigious CXR-LT 2026 Challenge, a competition dedicated to developing advanced AI methods for chest X-ray analysis.

The winning team includes Pham Ha Hieu and Nguyen Hai Dang, members of the Computer Vision and Medical AI Lab (CVMAIL), VinUni-Illinois Smart Health Center (VISHC), in collaboration with the Machine and Hybrid Intelligence Lab (MIHL) led by Dr. Ulas Bagci at Northwestern University.

The challenge is organized in conjunction with IEEE ISBI 2026 (International Symposium on Biomedical Imaging and supported by the US National Science Foundation Career Grant (2145640). The team’s solution achieved the highest ranking among 72 participating teams.

The CVMAIL × MIHL team ranked #1 in both competition tracks:

• Task 1 – Long-tailed Multi-label Classification
• Task 2 – Zero-shot / Out-of-Distribution (OOD) Multi-label Classification

The CXR-LT Challenge focuses on advancing AI methods for chest X-ray disease recognition under long-tailed data distributions and zero-shot settings. These challenges reflect real-world clinical scenarios where many diseases appear infrequently in training data, while AI systems must also be capable of generalizing to previously unseen conditions or distribution shifts. Developing robust solutions for such settings is essential for improving the reliability and clinical applicability of medical imaging AI.

Image from ISBI 2026: Congratulations to all members involved in this work: Ha Hieu Pham, Nguyen Hai Dang, Nguyen Thanh Huy, Assoc. Prof. Min Xu (CMU), Assoc. Prof. Ulas Bagci (NU), Dr. Le Trung Nghia (HCMUS), and Asst. Prof. Pham Huy Hieu. Special congratulations to Ha Hieu Pham, Research Assistant, and Nguyen Hai Dang, PhD student of the lab, who represented CVMAIL in the challenge and contributed significantly to this outstanding achievement.

For more information:

• Our technical paper: https://arxiv.org/pdf/2602.13430
• Original paper: https://arxiv.org/pdf/2602.22092
• Technical paper: https://lnkd.in/gc7cX-Vr