-
In this project, I created a 3D Segmentation Model to identify Brain Tumor.
-
Brain MRIs are in 3D volumes so it is very difficult to pass the whole 3D volume at a time. To overcome this issue I divided 3d volumes into 3D subvolumes and selected only subvolumes that have at least 5% Tumor area.
-
To create a Tensor dataset in I created a notebook in which MRI images which are in .nii.gz forms are converted and pre-processed(sub-volumes and Standardization) and then saved for training.
- Due to the lack of Computional Resources, I didn't use whole dataset.