Joshua Scheuplein MT final talk: A Foundation Model for X-Ray Imaging

Join us for the final presentation of a Master’s thesis that investigates the development of a versatile foundation model for medical X-ray image analysis. By adapting the self-supervised learning method DINO, the thesis explores how a large dataset of 632,000 unlabeled X-ray images can be used to train foundational backbones capable of extracting meaningful features for tasks like body region classification, metal implant segmentation, and screw object detection. The findings highlight the potential of foundation models to improve feature extraction in medical X-ray imaging while streamlining development processes at the same time.