Diffusion tensor subspace imaging of double diffusion–encoded MRI delineates small fibers and gray-matter microstructure not visible with single encoding techniques

Purpose: Double diffusion encoding (DDE) acquisition strategies promise specificity for small-dimensional structures inaccessible to single diffusion encoding (SDE). For DDE-weighted MRI scans to become relevant for whole brain imaging, signal reconstruction frameworks must accurately report microstructural features of interest—especially microscale anisotropy in complex tissue environments. This study examined the recently developed diffusion tensor subspace imaging (DiTSI) framework and its radial and spherical anisotropy metrics (RA and SA, respectively) in postmortem human brain tissue specimens. Methods: MRI microscopy including multishell SDE-weighted and DDE-weighted imaging was performed for healthy brain stem and temporal lobe specimens and for specimens with Alzheimer's disease pathology and neurodegeneration. The DiTSI framework was compared with four other diffusion MRI frameworks, and angular and radial DDE sampling were evaluated. Results: DDE acquisition and the DiTSI metric maps of SA and RA in temporal lobe and brain-stem specimens were found to be distinct from fractional anisotropy and orientation dispersion index in providing complementary and selective contrast of microscale anisotropy at the gray-matter/white-matter interface in the cortex and in hippocampal layers. DiTSI maps also unmasked small fascicles in the brain stem that were not detectable by SDE techniques and provided selective contrast across the major fiber pathways. Results also revealed prominent reductions of SA and RA in tissue with Alzheimer's disease pathology that were not observed for any other framework. Conclusions: New contrasts were evident for DiTSI framework metrics over a range of tissue environments with promise toward providing novel markers of pathology.