niftytorchprep

BIDS to NiftyTorch: data preparation for deep learning.

How it works?

niftytorchprep command-line interface tool helps you prepare your data for niftytorch training. It check if BIDS format of your data is correct and transforms it into the format that is coherent with deep learning models training.

Install

pip install git+https://github.com/NiftyTorch/ohbm-hackthon2020.git

Or download this repository and call:

python setup.py install

Interface

$ niftytorchprep --help  
Usage: niftytorchprep [OPTIONS] COMMAND [ARGS]...

  NIFTYTORCHPREP helps to get your data ready for *niftytorch* training. You
  can browse through your options below. Each one has respective help
  function.

Options:
  --help  Show this message and exit.

Commands:
  bids-files       Print types of files and their number per folder.
  bids-totraining  Takes data from BIDS_DIR and organises it in a training...
  bids-validate    Basic BIDS verification.
  qc-anat          Runs Quality Control (T1) from visualqc package and...
  qc-func          Runs Quality Control (Functional) from visualqc package...
  qc-getvisualqc   Installs visualqc from PIP.

Here is an example that splits the data from various participants in BIDS format among: training, test and validation folders:

niftytorchprep bids-totraining my_bids_data/ my_output_to_dl/ gender --test 0.2 --val 0.1

To get help for a specific option, call for example:

niftytorchprep bids-totraining --help

See it in action:

Contributors

Dominik Krzemiński @dokato
Cardiff University Brain Research Imaging Centre

Sara Morsy @SaraMorsy
Faculty of Medicine, Tanta University, Egypt

Kaori Lily Ito @kaoriito
Neural Plasticity & Neurorehabilitaiton Laboratory, University of Southern California

History

This project has been initiated at OHBM Hackthon 2020.

Haoyulance / niftytorchprep