/VoxResNet

An implementation of VoxResNet for brain segmentation.

Primary LanguagePython

VoxResNet

This is an unofficial implementation of VoxResNet: Deep Voxelwise Residual Networks for Volumetric Brain Segmentation.

Requirements

  • python 3.6
  • TensorFlow
  • TensorLayer
  • NiBabel
  • Numpy
  • SciPy

Data Setup

  1. Download ADNI dataset from http://adni.loni.usc.edu. (or other dataset you want)
  2. Modify file path and run preprocess.py.
    • You can skip this step if data augmentation is not required.
    • And make sure all of your images have the same size.

Training

  1. Follow steps in the Data Setup
  2. Run train.py

Testing

  1. Follow steps in the Data Setup
  2. Run segment.py, and the outputs are saved to save_path
  3. Run dice.py, dice is defined as: