Document Type

Thesis

Degree

Master of Arts

Major

Psychology

Date of Defense

4-20-2020

Graduate Advisor

Carissa L. Philippi

Committee

Sandra Langeslag

John Meriac

Abstract

Clinical research has revealed aberrant activity and connectivity in default mode (DMN), frontoparietal (FPN), and salience (SN) network regions in major depressive disorder (MDD). Recent functional magnetic resonance imaging (fMRI) studies suggest that variability in brain activity, or blood oxygen level-dependent (BOLD) signal variability, may be an important novel predictor of psychopathology. However, to our knowledge, no studies have yet determined the relationship between resting-state BOLD signal variability and MDD nor applied BOLD signal variability features to the classification of MDD history using machine learning (ML). Thus, the current study had three aims: (i) to investigate the differences in the voxel-wise resting-state BOLD signal variability between varying depression histories; (ii) to examine the relationship between depressive symptom severity and resting-state BOLD signal variability; (iii) to explore the capability of resting-state BOLD signal variability to classify individuals by depression history. Using resting-state neuroimaging data for 79 women collected as a part of a larger NIH R01-funded study, we conducted (i) a one-way between-subjects ANCOVA, (ii) a multivariate multiple regression, and (iii) applied BOLD signal variability and average BOLD signal features to a supervised ML model. First, results indicated that individuals with any history of depression had significantly decreased BOLD signal variability in the left and right cerebellum and right parietal cortex in comparison to those with no depression history (pFWE < .05). Second, and consistent with the results for depression history, depression severity was associated with reduced BOLD signal variability in the cerebellum. Lastly, a random forest model classified participant depression history with 76% accuracy, with BOLD signal variability features showing greater discriminative power than average BOLD signal features. These findings provide support for resting-state BOLD signal variability as a novel marker of neural dysfunction and implicate decreased neural signal variability as a neurobiological mechanism of depression.

Download

Share

COinS