kulendu/DeepSeg3D

This repository contains a TensorFlow implementation of Semantic Segmentation of Brain Tumor on multi-band 3D volumes using non-uniform 3D U-Nets

This repository contains a TensorFlow implementation of Semantic Segmentation of Brain Tumor on multi-band 3D volumes using non-uniform 3D UNets

This codebase provides:

• Testing code
• Visualization code

Requirements

• Python 3
• TensorFlow tested on 2.4.1
• os
• numpy
• tqdm
• Keras
• glob
• Segmentation Models 3D

Run the following code to install all pip packages:

python3 ./scripts/pkg_install.py

Prepare the data

1. run the following script to download the generated and conerted data (in pkl format) and copy them under the root folder ./

sh ./scripts/download_generated_pkl_files.sh 

2. run the following script to download the full dataset

sh ./scripts/download_full_data.sh 

Architecture definations

The model definations are shown here:

Figure: Detailed parameters for all the five models

w/ all the model's performance:

Figure: Info. and statistical data of the models