Not Logged In

Learning to Classify Psychiatric Disorders based on fMR Images: Autism vs Healthy and ADHD vs Healthy

Full Text: MLINIFinal.pdf PDF

A clinical tool that can diagnose psychiatric illness using functional magnetic resonance brain images would greatly assist physicians. Here, we propose a learning algorithm that uses the histogram of oriented gradients (HOG) features of fMRI images as the input to support vector machines, then show that this system can produce such classi ers when run on two large public datasets: able to diagnose ADHD with hold-out accuracy of 0:626 (over baseline = 0:550) when trained on the ADHD-200 global competition dataset, and to diagnose autism with hold-out accuracy of 0:619 (over baseline = 0:516) when trained on the Autism Brain Imaging Data Exchange (ABIDE) dataset. While these results are not yet to clinical relevance, they outperform all previously presented methods on both datasets. These results suggest that our learning approach may lead to diagnostic classi ers (from functional images) for yet other psychiatric disorders.

Citation

S. Ghiassian, R. Greiner, M. Brown, P. Jin. "Learning to Classify Psychiatric Disorders based on fMR Images: Autism vs Healthy and ADHD vs Healthy". Proceedings of the Workshop on Machine Learning and Interpretation in Neuroimaging, pp n/a, December 2013.

Keywords: FMRI, Machine Learning, Classification, ADHD, Autism, ADHD-200, HOG Features, computational psychiatry
Category: In Workshop
Web Links: MLINI

BibTeX

@misc{Ghiassian+al:MLINI13,
  author = {Sina Ghiassian and Russ Greiner and Matt Brown and Ping Jin},
  title = {Learning to Classify Psychiatric Disorders based on fMR Images:
    Autism vs Healthy and ADHD vs Healthy},
  Pages = {n/a},
  booktitle = {Proceedings of the Workshop on Machine Learning and
    Interpretation in Neuroimaging},
  year = 2013,
}

Last Updated: February 12, 2020
Submitted by Sabina P

University of Alberta Logo AICML Logo