Abstract
A mathematical model-based parcellation of magnetic resonance diffusion tensor images (DTI) has been developed to quantify progressive changes in three types of tissues - grey (GM), white matter (WM), and damaged spinal cord tissue, along with behavioral assessments over a 6 month period following targeted spinal cord injuries (SCI) in monkeys. Sigmoid Gompertz function based fittings of DTI metrics provide early indicators that correlate with, and predict, recovery of hand grasping behavior. Our three tissue pool model provided unbiased, data-driven segmentation of spinal cord images and identified DTI metrics that can serve as reliable biomarkers of severity of spinal cord injuries and predictors of behavioral outcomes.