My research addresses the burden on radiologists by integrating a “radiologist-in-the-loop” approach to improve automated medical image interpretation. With a shortage of qualified radiologists, there’s an urgent need for tools that streamline routine case analysis while supporting complex diagnoses. I’m developing an interactive machine learning (IML) framework with Active Learning (AL) that allows radiologists to actively participate in the model’s training. By applying decision rules and focusing on key image sections (prototypical parts), radiologists can guide the model’s learning. This approach enhances diagnostic accuracy, effectively handles diverse cases, and builds trust through clear, interpretable explanations.