White matter tractography mapping relies on accurate manual delineation of anatomical seeding regions of interest (sROIs) by neuroanatomy experts, but stringent pre-operative time constraints and limited availability of experts create a critical bottleneck for neuro-surgical planning and navigation. Moreover, the manual segmentation process is labor-intensive and not feasible given the time pressures of pre-surgical planning, particularly for tumor resection candidates who require rapid assessment.

Technology
• Multi-modal fully convolutional neural networks for automated sROI segmentation. The technology uses deep learning architectures that fuse anatomical information from T1-weighted (T1w) MRI maps with directionally encoded color (DEC) maps derived from diffusion imaging to automatically segment seeding ROIs.
• Mimics expert practice through data fusion. Inspired by how neuroanatomy experts manually perform segmentation, the networks integrate multiple imaging modalities to compute accurate segmentations of critical white matter tract starting points.
• Targets three critical pathways. The system segments sROIs for the motor tract, arcuate fasciculus, and optic radiation—all crucial structures for surgical planning to avoid neurological deficits.
The resulting system is a scalable solution for expert shortage. It dramatically improves efficiency without quality compromise, is validated on clinical data and addresses critical clinical workflow constraint. By automating what was previously a manual, expert-dependent process, the technology directly solves the time-constraint problem in pre-operative planning where rapid turnaround is essential.

Potential Applications
• Pre-surgical planning for brain tumor resection. The primary application is automating tractography mapping for the 75 real tumor resection candidates studied, enabling faster and more consistent surgical planning.
• Neuro-surgical navigation support. Automated sROI segmentation can improve intraoperative navigation systems by providing reliable white matter tract maps without requiring expert manual delineation during time-critical pre-operative periods.
• Standardization across institutions. By removing inter-rater variability inherent in manual segmentation, the technology enables more consistent tractography results across different hospitals and surgical teams.
• Research and clinical trials. The automated approach facilitates larger-scale studies of white matter pathways by eliminating the expert labor bottleneck.

Patents and articles supporting the technology:
• Patent US11170508B2 granted
• Avital et al 2020, IEEE TRANSACTIONS ON MEDICAL IMAGING