This book tackles a key problem in brain cancer diagnosis by introducing a refined version of the well-known U-Net model. The method extends the architecture developed by M. Merati, A. Z. Lebani, and S. Mahmoudi, which reached a Dice score of 95.82% and an IoU of 99.61%. These results surpass the standard U-Net and other reported models.
Part I: Introduction to the medical and technical context
Chapter 1: Clinical Challenges of Brain Tumor Segmentation
Chapter 2: State of the art in medical image segmentation
Part II: Methodological Foundations
Chapter 3: Data Preparation and Augmentation