Poster B109, Sunday, March 26, 8:00 – 10:00 am, Pacific Concourse
Using Patterns of Functional Brain Connectivity to Predict Autism Spectrum Disorder
Hakeem Brooks1, Jin Cheong2, Jeremy Cohen1, Luke Chang2; 1Xavier University of Louisiana, New Orleans, LA, 2Dartmouth College, Hanover, NH
Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by social and communication deficits and repetitive behaviors. Considerable work has focused on characterizing the neurobiological sequelae of this disorder, but no reliable pathognomonic biological marker of this disease has yet emerged. One potential reason is that most studies use relatively modest sample sizes and are focused with identifying correlates of the disorder rather than predictive markers. To address these limitations, we combined multivariate statistical learning techniques with a large open access resting state functional magnetic resonance imaging dataset (Autism Brain Imaging Data Exchange II; ABIDE II) to develop predictive brain markers of ASD. We used functional connectivity metrics to measure how well different regions of the brain communicate with each other as features in our predictive model. Our classifier was able to consistently distinguish between ASD and neurotypical populations groups and revealed which specific brain connections most reliably contribute to discriminating between the two groups. Our results demonstrated that functional neuroimaging, and functional connectivity specifically, could provide a viable approach to quantifying aspects of ASD etiology.
Topic Area: NEUROANATOMY