Date of Award
3-24-2016
Document Type
Thesis
Degree Name
Master of Science
Department
Department of Mathematics and Statistics
First Advisor
Ryan D. Kappedal, PhD.
Abstract
Autism spectrum disorder (ASD) is a neurodevelopmental condition that can be debilitating to social functioning. Previous functional Magnetic Resonance Imaging (fMRI) classification studies have included only small subject sample sizes (n 50) and have seen high classification accuracy. The recent release of the Autism Brain Imaging Data Exchange (ABIDE) provides fMRI data for over 1,100 subjects. In our research, we derive a subject's functional network connectivity (FNC) from their fMRI data and develop a regularized logistic classifier to determine whether a subject has autism. We obtained up to 65% classification accuracy, similar to other studies using the ABIDE dataset, suggesting that generalizing a classifier over a large number of subjects is much more difficult than smaller studies. The connectivity among several brain regions of ASD subjects were highlighted in the model as abnormal compared to the control subjects which potentially warrants future investigations about how these regions affect ASD. Although the classification accuracy was lower than what could be considered as clinically applicable, this research contributes to the continuing development of an automated classifier for diagnosing autism.
AFIT Designator
AFIT-ENC-MS-16-M-123
DTIC Accession Number
AD1008556
Recommended Citation
Palko, Kyle A., "Diagnosing Autism Spectrum Disorder through Brain Functional Magnetic Resonance Imaging" (2016). Theses and Dissertations. 248.
https://scholar.afit.edu/etd/248