Differences in resting state and task-based EEG measures between patients with major depressive disorder and healthy controls

Abstract

Objective: Assessing depression in psychiatry relies on subjective measures that may not adequately reflect the disorder's biology. Electroencephalography offers an objective and scalable approach for gathering data with potential for characterizing major depressive disorder. We explore the potential of a combination of EEG-based neurocognitive measures for the characterization of depression.

Methods: Resting state measures and electrophysiological responses during emotional faces recognition and a three-choice vigilance task, were examined in a sample of depressed patients and healthy controls.

Results: The findings revealed differences in resting state spectral power measures in the theta, alpha, and beta ranges. Relative alpha power in eyes open condition was decreased in patients and the degree of reduction was correlated with the severity of both anxiety and depressive symptoms. The N170 face component of the evoked responses to emotional faces captured depression-related emotional bias towards sad faces. The three-choice vigilance task demonstrated depression-related attentional behavioral deficits, and an increase in P200 amplitude which was also associated with greater depression severity.

Conclusions: The three paradigms revealed distinct and complementary EEG signatures of depression.

Significance: Our findings suggest the benefits of utilizing objective measures for enhancing our understanding and treatment of the disorder.

Keywords: Anxiety; Depression; EEG; ERP; Electroencephalogram; Major depressive disorder; N170; Power spectrum; Signature.

Fig. 1.

Emotional image recognition task paradigm. During the encoding phase Happy, Sad, and Neutral faces are presented in random order. In the test phase, the same stimuli are mixed with novel faces and participants are asked to indicate previously seen faces. The images presented in the figure were sourced from the publicly available subset of the FACSES database, approved for research publication reproduction (source: https://faces.mpdl.mpg.de/imeji/).

Fig. 2.

Three-choice vigilance task (3CVT). Three different geometrical shapes (including Target, NonTarget, and Distractor) appear individually at random locations on the screen in a randomized order and with widely ranging inter-stimulus intervals (ISI).

Fig. 3.

Resting state signatures of depression. A. Relative power spectra in MDD and HC groups at the central electrode in Eyes Open condition. Beta range 15–40 Hz is displayed in the original and enlarged scales. B. Scalp maps for relative Alpha power in MDD and HC, and statistical results of the differences between the groups. The last scalp map represents normalized effect sizes (ES) across channels, overlayed with markers of significance. C. Scatterplots of correlations between average relative Alpha power and clinical scores, with outliers marked with red crosses.

Fig. 4.

N170 emotional processing bias in patients with MDD and healthy controls. A. Group average ERPs at channel Cz for Sad and Neutral Target conditions. The pink rectangle highlights the time window for N170 peak measurement. B. Spatial distributions of the N170 area under the curve (AUC) for Sad target stimuli, Neutral target stimuli, and the difference between Sad and Neutral stimuli along with the corresponding effect sizes are shown for the MDD and HC groups. Statistically significant within group differences in AUC for Sad and Neutral stimuli are shown in the third column for MDD and HC groups.

Fig. 5.

P200 mean amplitude from the three-choice vigilance task. A. Group average ERPs for Target condition at channel Fz. The P200 is a positive ERP component that peaks after 200ms (red arrow). The P200 mean amplitude is calculated by averaging ERP amplitudes within the 190–300ms time window (highlighted in pink). B. Spatial distributions of the mean P200 amplitude, across the HC and MDD groups. The rightmost scalp map displays the effect sizes of the HC-MDD differences, with statistical significance indicated by overlaid markers. C. Scatterplots showing correlations between univariate average P200 amplitudes and clinical scores, with outliers marked with red crosses.

Fig. 6.

Correlation analysis between univariate depression signatures from resting state, emotional image recognition task, and three-choice vigilance task. Additionally, clinical scores are presented on the right side of the vertical dotted line and below the horizontal dotted line to demonstrate each depression signature in relation to severity of depression and anxiety. Asterisks (*) indicate significant correlations at p<0.05.