Format

Send to

Choose Destination
Bipolar Disord. 2016 Nov;18(7):612-623. doi: 10.1111/bdi.12446. Epub 2016 Nov 5.

Distinguishing medication-free subjects with unipolar disorder from subjects with bipolar disorder: state matters.

Author information

1
Program for Mood Disorders, Department of Psychiatry, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands.
2
Department of Psychiatry, University of Münster, Münster, Germany.
3
Department of Psychiatry and Neuroscience, Campus Amsterdam, VU University Medical Center, Amsterdam, the Netherlands.
4
Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands.
5
Department of Psychiatry, VU University Medical Center, Amsterdam, the Netherlands.
6
Department of Psychiatry, Radboud University Medical Center, Nijmegen, the Netherlands.
7
Department of Psychiatry, Mood and Anxiety Disorders, University of Groningen, University Medical Center Groningen, Groningen, the Netherlands.

Abstract

OBJECTIVES:

Recent studies have indicated that pattern recognition techniques of functional magnetic resonance imaging (fMRI) data for individual classification may be valuable for distinguishing between major depressive disorder (MDD) and bipolar disorder (BD). Importantly, medication may have affected previous classification results as subjects with MDD and BD use different classes of medication. Furthermore, almost all studies have investigated only depressed subjects. Therefore, we focused on medication-free subjects. We additionally investigated whether classification would be mood state independent by including depressed and remitted subjects alike.

METHODS:

We applied Gaussian process classifiers to investigate the discriminatory power of structural MRI (gray matter volumes of emotion regulation areas) and resting-state fMRI (resting-state networks implicated in mood disorders: default mode network [DMN], salience network [SN], and lateralized frontoparietal networks [FPNs]) in depressed (n=42) and remitted (n=49) medication-free subjects with MDD and BD.

RESULTS:

Depressed subjects with MDD and BD could be classified based on the gray matter volumes of emotion regulation areas as well as DMN functional connectivity with 69.1% prediction accuracy. Prediction accuracy using the FPNs and SN did not exceed chance level. It was not possible to discriminate between remitted subjects with MDD and BD.

CONCLUSIONS:

For the first time, we showed that medication-free subjects with MDD and BD can be differentiated based on structural MRI as well as resting-state functional connectivity. Importantly, the results indicated that research concerning diagnostic neuroimaging tools distinguishing between MDD and BD should consider mood state as only depressed subjects with MDD and BD could be correctly classified. Future studies, in larger samples are needed to investigate whether the results can be generalized to medication-naïve or first-episode subjects.

KEYWORDS:

fMRI ; bipolar disorder; diagnosis; machine learning; major depressive disorder; mood state

PMID:
27870505
DOI:
10.1111/bdi.12446
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Wiley
Loading ...
Support Center