Assessment, identification, and classification of cortical abnormalities in autistic adults using diffusion tensor imaging of underlying white matter

Autism is a neurodevelopmental disorder characterized by communication deficits, abnormal social interactions, and restrictive or repetitive interests and behaviors. Although autism research is an emerging field, the cause and cure remain unknown. Previous studies suggest an abnormal brain growth pattern in autism with near normal brain volume at birth, enlargement during early childhood, slowed growth during adolescence, and again, near normal brain volume during adulthood. Early brain growth abnormalities may negatively affect maturing brain areas. This leads to microscopic differences in white matter tract structure and organization which is undetectable by traditional MRI methods. By focusing on the less studied adult autistic population, the current diffusion tensor imaging study uses brain water diffusion to probe tissue organization and assess structural alterations resulting from an abnormal brain growth pattern.; Typical diffusion imaging studies focus on central white matter regions to evaluate damage. The current study maps damage throughout the entire cerebral cortex using a post-processing method. In this method, subcortical diffusivity measures from white matter directly underlying the cortex are mapped to the overlying cortex. These interface maps were thoroughly investigated through a region of interest (ROI) and a voxelwise approach. Both analyses provide complementary results, finding similar areas of increased diffusivity in autistic individuals, such as middle temporal and inferior frontal areas. However, each analysis also reports several unique results. In the ROI case, autistic individuals additionally display significantly increased diffusivity in the left inferior temporal region and bilateral occipital cortex. In the voxelwise case, autistic individuals additionally display significantly increased diffusivity in the left inferior parietal lobule, bilateral precuneus, and bilateral insula. A subsequent surface-based analysis allowed visualization of these differences on the entire three dimensional cortical surface. The apparent differences between analyses are due to their inherent methodology. ROI analysis identifies larger diffusivity changes and is capable of detecting areas that may vary among subjects in their anatomical location within a region, while the voxelwise analysis finds smaller areas of change. Importantly, all identified areas are implicated in autism. Interestingly, the analyses collectively report more mean diffusivity (MD), parallel diffusivity (lambda||), and perpendicular diffusivity (lambda⊥) differences compared to fractional anisotropy (FA) for autistic individuals, suggesting greater MD sensitivity to diffusivity differences at the gray/white matter interface. A simple diffusion simulation was carried out to investigate relative FA and MD changes for slight modifications of parallel axonal tract organization. Due to greater FA variability and similar relative FA and MD changes, results indicate that MD is a more sensitive measure, further supporting the previous ROI, voxelwise, and surface analyses' findings.; Interface maps are potential tools for clinically diagnosing autism. Autistic and control interface diffusion data was classified using two different automated techniques, the Earth Mover's Distance Metric and the Center of Mass. Initial investigation on the small sample size yields promising results with 13 out of 16 individuals classified correctly. This suggests the use of interface maps as biomarkers in the automated diagnosis and classification of autism.