segmentation-project.github.io - Semantic Segmentation of Brain Tumor on multi-band 3D volumes using non-uniform 3D U-Net

Description: Deformable Neural Radiance Fields creates free-viewpoint portraits (nerfies) from casually captured videos.

nerf (171) d-nerf (70) nerfies (69)

Example domain paragraphs

A brain tumor is a cancerous and non-cancerous mass or growth of abnormal cells in the brain. It can begin else- where and spread to the brain. There is considerable significance in MR-Images of the brain in identifying the outline of the tumor and in identifying clinical relevance in the diagnosis, prognosis, and treatment of the tumor.

Recent improvements using deep learning models have proved their effectiveness in various seg- mentation and medical imaging tasks, many of which are based on the U-Net network structure with symmetric encoding and decoding paths for end-to-end segmentation.

In this work, we aim to develop a pipeline consisting of a baseline deep learning model with 3D U-Net constituting adaptation in the training, model structure, and model parameters/hyper-parameters for semantic segmentation of brain tumors. Furthermore, instead of using one model for best results, multiple variants of the U-Net were trained with tweaked hyper-parameters and encoding/decoding blocks to reduce errors and improve performance. Brain Tumor Segmentation (BraTS) Challenge 2020 data was chosen as t

Links to segmentation-project.github.io (1)