Glioblastoma 3D Segmentation

The goal of this project is to train an accurate 3D Neural Network for Glioblastoma Tumor Segmentation using a limited dataset. To this end, we make use of Patch-Learning. Our work combines the BraTS21 winning U-Net design : Optimized U-Net for Brain Tumor Segmentation and the BraTS20 winning learning approach : nnU-Net for Brain Tumor Segmentation.