Structural MRI-based discrimination between autistic and typically developing brain

Autism is a neurodevelopmental disorder characterized by marked deficits in communication, social interaction, and interests. Various studies of autism have suggested abnormalities in several brain regions, with an increasing agreement on the abnormal anatomy of the white matter (WM) and on deficits in the size of the corpus callosum (CC) and its sub-regions in autism. In this paper, we aim at using these abnormalities in order to devise robust classification methods of autistic vs. typically developing brains by analyzing their respective MRIs. Our analysis is based on shape descriptions and geometric models. We compute the 3D distance map to describe the shape of the WM, and use it as a statistical feature to discriminate between the two groups. We also use our recently proposed non-rigid registration technique to devise another classification approach by statistically analyzing and comparing the deformation fields generated from registering segmented CC's onto each others. The proposed techniques are tested on postmortem and on in-vivo brain MR data. At the 85% confidence level the WM-based classification algorithm correctly classified 14/14 postmortem-autistics and 12/12 in-vivo autistics, a 100% accuracy rate, and 13/15 postmortem controls (86% accuracy rate) and 30/30 in-vivo controls (100% accuracy rate). The technique based on the analysis of the CC was applied only on the in vivo data. At the 85% confidence rate, this technique correctly classified 10/15 autistics, a 0.66 accuracy rate, and 29/30 controls, a 0.96 accuracy rate. These results are very promising and show that, contrary to traditional methods, the proposed techniques are less sensitive to age and volume effects.