Early Prediction of Autism Spectrum Disorder at 12 months of Age from Brain Morphometry

Public Deposited

Analytics

Last Modified

Creator

D. Ben Knoble
- Affiliation: College of Arts and Sciences, Department of Computer Science

Abstract

Recent research established a link between the development of autism spectrum disorder (ASD) and increased levels of extra-axial cerebrospinal fluid (EA-CSF) [30]. Along with morphometric measures, local measures such as EA-CSF enlargement have also been shown to be relevant in a classification model of ASD from brain magnetic resonance imaging (MRI) images of infants aged 6 months [22]. To analyze the predictive usefulness of the 12-month MRI data as compared to the 6-month data, I extracted local EA-CSF measures from 12-month MRIs in Infant Brain Imaging Study (IBIS) subjects. I then adapted the 6-month model for use with the 12-month measurements. While I hypothesized that the 12-month features would prove more predictive than the 6-month features, my cur- rent findings are that for almost all brain morphometric measurements, the adapted classifiers performed more weakly on the 12-month data set than on the 6 month data. I explore some possible underlying reasons in this study. Further research is needed, however, to determine whether a combination of brain morphometric measurements or additional classi- fier optimization may lead to a sufficiently accurate prediction for clinical predictive use.

Date of publication

Keyword

Subject

DOI

Resource type

Rights statement

License

Advisor

Martin A. Styner
- Affiliation: College of Arts and Sciences, Department of Computer Science
- Other Affiliation: Neuro Image Research and Analysis Laboratories

Degree

Graduation year

Language

Relations

Parents:

This work has no parents.

In Collection:

Thumbnail	Title	Date Uploaded	Visibility	Actions
	thesis.pdf	2020-04-30	Public	Download