Incorporating Task-Specific Structural Knowledge into CNNs for Brain Midline Shift Detection

[paper]

This repository contains full code for training and inference of the model described in our paper.

Install

git clone https://github.com/neuro-ml/midline-shift-detection
cd midline-shift-detection
pip install -r requirements.txt
pip install -e .

Inference

python scripts/predict.py IMAGE OUTPUT_CONTOURS
# or run 
python scripts/predict.py --help
# for more details

All the images must be .nii or .nii.gz files containing axial MRI series.

Training

python scripts/train.py DATA OUTPUT_MODEL
# or run
python scripts/train.py --help
# for more details

DATA is a folder containing the training set with the following structure:

DATA:
    - filename1.nii.gz
    - filename1.json
    - filename2.nii.gz
    - filename2.json
    ...

where the json files contain a list of annotations for a given image. See the data folder for an example.