This project implements a 3-layer U-Net model in PyTorch to perform brain tumor segmentation on the LGG MRI dataset. The model learns to predict binary segmentation masks (tumor vs. non-tumor regions) from MRI scans.