Tools
The software we build to measure brain physiology is released openly on GitHub, so other labs can reproduce our work and use it in their own studies.
VeinSeg: cerebral vein segmentation
Veins carry deoxygenated blood, so mapping them precisely is the first step to measuring how much oxygen the brain extracts (OEF). VeinSeg is a 3D deep-learning tool that segments cerebral veins from a QSM or R2* map.
- Trained on multi-site data at 3T and 7T, across five QSM reconstruction methods (TGV, MEDI, L1, STAR, iLSQR) and R2*
- Physics-informed: training combines Dice and cross-entropy with a dipole field-consistency loss and a Frangi vesselness shape loss
- Needs only a single image as input, and outputs a binary vein mask and a probability map
pip install veinseg-qsm
veinseg -i qsm.nii.gz -r tgv -f 7t -o mask.nii.gz -p prob.nii.gz
Coming soon
We are preparing more of our pipelines for release, including tools for mapping oxygen extraction (OEF) and oxygen metabolism (CMRO2), and for separating the magnetic signatures of veins, iron and myelin. Follow us on GitHub to hear when they are out.
