Detecting recent concussion with multidimensional diffusion imaging: enhanced sensitivity over fractional anisotropy

Despite its classification as a "mild" brain injury, concussion causes a cascade of potentially damaging neural effects. While not often visually detectable on clinical MRI scans, concussion-induced white matter (WM) injury can be detected by diffusion imaging using fractional anisotropy (FA). Though FA is interpreted as a marker of "WM integrity", it cannot distinguish WM microstructural changes from fiber directionality variation. This limitation highlights the need for alternative imaging biomarkers. Multidimensional Diffusion (MDM) imaging provides estimation of "micro" FA ({micro}FA), which is more robust to fiber directionality variation. Yet, it is unknown whether {micro}FA and other MDM parameters can reliably detect concussion-related injury. We employed MDM imaging to estimate diffusion markers across WM regions in Concussion (N=30, mean age=21.1, sex=13M/17F, scanned within 5 days of injury) and Control (N=32, mean age=22.8, sex=18M/14F) groups. We assessed: 1) whether regional MDM parameter values are credibly associated with concussion, and 2) whether MDM parameter group discriminability outperforms F A. Lower MDM anisotropy ({micro}FA, D2anison) and greater MDM isotropy (Diso, Vison; equivalents to mean diffusivity [MD] and MD variance) values were credibly associated with concussion across several regions (94% HDI for {beta} excluded 0). These trends were not unique to areas with crossing fibers. {micro}FA and D2anison displayed significantly greater discriminability (higher AUC) when compared to FA (p<0.001). MDM parameters, particularly {micro}FA and D2anison, are promising biomarkers of concussion-induced WM injury. The lack of specificity to crossing fiber areas challenges the idea that axonal shearing drives concussion-induced injury.