neurosainteanne/synthflair

Synthetic FLAIR generation software

synthFLAIR

Software designed to convert MRI diffusion-weighted sequences into FLAIR sequences

Prerequisites

Python 3.8

Usage

First, clone the git directory :

git clone https://github.com/NeuroSainteAnne/synthFLAIR.git
cd synthFLAIR/

Then install required modules:

pip install -r requirements.txt
pip install jupyterlab

Open the jupyter notebook

jupyter notebook

Data preparation

Open the synthFLAIR-Dataprep in order to convert you NIFTI files in a Memmap format for model training

Model training

Open the synthFLAIR-Train notebook and follow instructions to train and save your model.

Alternatively, download the pre-trained weights available here.

Inference

Once the model is trained and saved, you can convert 4D NIFTI DWI volumes into synthFLAIR volumes with the synthFLAIR-Convert tool.

Orthanc module

Once the model is trained and saved, you can use the Orthanc module to automatically convert DWI volumes into FLAIR using a lightweight DICOM server. Follow the intructions at Orthanc Module Readme