Anatomy to Tract Mapping: A Repeatability Study

Motivation: Anatomy to Tract Mapping (ATM) has shown promising performance in generating bundle-specific streamlines using only anatomical images. To ensure repeatability, we evaluate the test-retest reliability of ATM in reconstructing major white matter bundles. Goal(s): To evaluate ATM repeatability in bundle-specific streamline generation from test-retest anatomical MR images. Approach: We assess the reliability of ATM in producing consistent streamlines across test-retest sessions for various white matter bundles. Results: ATM demonstrates strong consistency in generating complete bundles for test-retest subjects, indicating high repeatability. It shows substantial agreement in bundle similarity, coverage, streamline validity, geometry and connectomic metrics. Impact: We show that bundle streamlines can be estimated directly from anatomical MRI. The generated streamlines exhibit robust test-retest reliability, contributing to (1) inferring streamlines when diffusion MRI is unavailable, (2) reducing local uncertainties in tractography, enhancing white matter pathways reconstruction.