About Brain Regressor CNN

A deep learning-based method to estimate brain morphometry directly from T1-weighted MRI.

The code is related to the following publication publication (bibtex):

Rebsamen, M., Suter, Y., Wiest, R., Reyes, M., & Rummel, C. (2020)
Brain morphometry estimation: From hours to seconds using deep learning.
Frontiers in Neurology, 11, 244. https://doi.org/10.3389/fneur.2020.00244

Quick start in three steps:

Create virtual environment (optional)

conda create -n DL python=3.6
source activate DL
pip install -r Python/requirements.txt

Import FreeSurfer data into h5 dataset

Assuming FS_RESULTS_DIR points to subjects directory with FreeSurfer results (GT) to import.

cd Python
./create_ds.sh FS_RESULTS_DIR

Train model

export PYTHONPATH=../miapy:../Python
python main.py config.json