Format

Send to

Choose Destination
PLoS One. 2016 Mar 14;11(3):e0151391. doi: 10.1371/journal.pone.0151391. eCollection 2016.

Seeking Optimal Region-Of-Interest (ROI) Single-Value Summary Measures for fMRI Studies in Imaging Genetics.

Author information

1
Clinical and Translational Neuroscience Branch, Division of Intramural Research Programs, National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland, United States of America.
2
Lieber Institute for Brain Development, Johns Hopkins Medical Campus, Baltimore, Maryland, United States of America.
3
Department of Statistics, University of Warwick, Warwick, United Kingdom.
4
Department of Psychiatry, University of Pennsylvania, Philadelphia, Pennsylvania, United States of America.
5
Departments of Psychiatry, Neurology and Neuroscience, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States of America.
6
Departments of Neurology and Radiology, Johns Hopkins University School of Medicine, Baltimore, Maryland, United States of America.

Abstract

A data-driven hypothesis-free genome-wide association (GWA) approach in imaging genetics studies allows screening the entire genome to discover novel genes that modulate brain structure, chemistry, and function. However, a whole brain voxel-wise analysis approach in such genome-wide based imaging genetic studies can be computationally intense and also likely has low statistical power since a stringent multiple comparisons correction is needed for searching over the entire genome and brain. In imaging genetics with functional magnetic resonance imaging (fMRI) phenotypes, since many experimental paradigms activate focal regions that can be pre-specified based on a priori knowledge, reducing the voxel-wise search to single-value summary measures within a priori ROIs could prove efficient and promising. The goal of this investigation is to evaluate the sensitivity and reliability of different single-value ROI summary measures and provide guidance in future work. Four different fMRI databases were tested and comparisons across different groups (patients with schizophrenia, their siblings, vs. normal control subjects; across genotype groups) were conducted. Our results show that four of these measures, particularly those that represent values from the top most-activated voxels within an ROI are more powerful at reliably detecting group differences and generating greater effect sizes than the others.

PMID:
26974435
PMCID:
PMC4790904
DOI:
10.1371/journal.pone.0151391
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Public Library of Science Icon for PubMed Central
Loading ...
Support Center