Verified Novel Diffusion MRI Signatures of Neuroinflammation in Experimental Diffuse Traumatic Brain Injury

Abstract

Diffuse traumatic brain injury (TBI) represents the most common form of brain injury, yet conventional magnetic resonance imaging (MRI) fails to detect neuroinflammatory pathophysiology without overt structural abnormalities. Diffusion MRI (dMRI) offers sensitivity to microscale pathophysiological changes, but imaging signatures, characteristic patterns on scans, require validation against histopathological markers. This study established radiologic-pathologic correspondence between dMRI signatures and histopathologic targets in experimental diffuse TBI. Ex vivo dMRI was performed on adult Sprague Dawley rats (n = 4) 7 days post-injury using a 7T system with mean apparent propagator metrics including Return-to-Axis Probability (RTAP) and Return-to-Origin Probability (RTOP). Immunohistochemical analysis used markers for microglia (Iba1+, CD83), astrocytes (GFAP, AQP4), and neuronal integrity (MAP2). Receiver operating characteristic (ROC) analysis evaluated the diagnostic performance of each marker using skeleton analysis protocols. Quantitative morphological analysis compared the primary somatosensory barrel field cortex (S1BF) and the motor cortex (control). Spatial correspondence was established between dMRI hyperintensity in RTAP and RTOP maps, where qualitative analysis identified high concentrations of microglial (Iba1+) pathology in the S1BF. Iba1+ quantitative analysis revealed significantly lower microglial endpoints and higher process length in S1BF regions compared to motor cortex, representative of injured pathology. CD83 showed contrasting endpoint patterns to Iba1+ but similar process length increase, providing complementary validation of microglial activation. CD83 endpoints demonstrated the strongest diagnostic performance (AUC = 0.750), while Iba1+ endpoints revealed an inverse relationship requiring reinterpretation (AUC = 0.816). GFAP, AQP4, and MAP2 showed minimal regional differences and poor diagnostic performance, suggesting limited utility as biomarkers at this timepoint. Imaging confusion matrix analysis demonstrated perfect specificity in control regions but limited sensitivity (33.3%) in injury regions. This study provides verification of radiologic-pathologic correspondence between dMRI signatures and microglial activation in experimental diffuse TBI.