Abstract
PURPOSE: The aim of this work is to make the DTI along the perivascular space (DTI-ALPS-index) more robust with respect to region selection and the orientation of the head and fibers. We propose to address this matter by using the principal diffusion directions and an automated, atlas-based region of interest (ROI) placement. METHODS: Simulations were used to determine the dependence of the DTI-ALPS-index on the orientations of the head and nerve fibers. Human MRI was performed on 12 healthy volunteers at 3T using a 64 channel head coil (Cima.X Siemens), and the DTI-ALPS-index was calculated along the principal diffusion directions (ALPS-PAS) and along the field of view or laboratory frame (ALPS-LAB). To calculate the DTI-ALPS-index with reduced bias in native space, we developed a novel algorithm for an automatic ROI placement technique. Its calculated index results were compared to those obtained from a manual ROIplacement in native space and from an existing atlas-based ROI placement. Test-retest scans with varying head rotation were conducted for validation. RESULTS: Simulations showed that ALPS-PAS was more robust toward head and fiber rotations than ALPS-LAB. In vivo, ALPS-PAS yielded a 10% higher index than ALPS-LAB. The automated ROI placement led to a smaller difference in the DTI-ALPS-index between test and retest measurement compared to the manual ROI placement. Stability was enhanced with ALPS-PAS for all ROI placements and varying head rotation. CONCLUSION: Using the principal components of the diffusion and automated ROI selection increased the measured DTI-ALPS-index, improved the robustness toward head and fiber orientations and eliminated the need for manual region selection.