The MICCAI SIG for Biomedical Image Analysis Challenges (SIG-BIAC) is collaborating with Medical Imaging AGI's Last Exam (MEDAL) to create the world's most challenging and clinically impactful medical imaging AI benchmark.

MEDAL: A new benchmark for clinically relevant medical imaging A(G)I

Current benchmarks in medical image analysis often reflect what data is available, not necessarily what is clinically needed. MEDAL (Medical Imaging AGI's Last Exam) aims to change this by introducing a benchmark explicitly designed around clinically relevant, high-impact problems.

The goal is to create a rigorous evaluation framework for emerging medical imaging AGI that goes beyond narrow, task-specific performance. Instead, MEDAL focuses on assessing whether models can generalize across modalities, tasks, and domains-capabilities that are essential for real-world clinical use but largely untested in existing benchmarks.

To achieve this, MEDAL adopts a global, top-down approach: the community is invited to define and contribute the most important open challenges in medical imaging. A €1M incentive supports the contribution of high-quality, previously unpublished datasets, enabling a benchmark that is both diverse and clinically meaningful.

The project is led by Prof. Dr. Lena Maier-Hein, President of the MICCAI Special Interest Group for Challenges (SIG-BIAC). The SIG is closely involved in the project and represented on the MEDAL Advisory Board alongside international experts of various disciplines, including MICCAI's President Marius George Linguraru, ensuring strong alignment with the global research community.