Autism Spectrum Disorders (ASD) Characterization in Children by Decomposing MRI Brain Regions with Zernike Moments

摘要

Autism Spectrum Disorder (ASD) is a complex neurological condition characterized by a triad of signs: stereotyped behaviors, verbal and non-verbal communication problems and troubles in social interaction. The scientific community has been interested on quantifying anatomical brain alterations of this disorder to correlate the clinical signs with brain tissue changes. This work presents a fully automatic method to find out brain differences between patients diagnosed with autism and control subjects. After pre-processing, a template (MNI152) is registered to each evaluated brain, obtaining a set of segmented regions. Each region is mapped into a 2D collage image which is decomposed by the Zernike Moments, obtaining magnitude and phase. These features are then used to train, region per region, a binary SVM classifier. The method was evaluated in a children population, aged from 6 to 12 years, from the public database Autism Brain Imaging Data Exchange. The AUC values for the most representative brain region were 77% for ABIDE I and 76% for ABIDE II, demonstrating the robustness of the method.