Scientific Abstract

Background: Electroencephalography (EEG) studies indicate that, relative to healthy controls, adults with Major Depressive Disorder (MDD) have increased alpha (8-12Hz) power over frontal and posterior scalp. However, the retest reliability of frontal alpha has been questioned, and many studies still use tedious hand-scoring. This analysis evaluated a fully automated EEG processing pipeline and investigated the reliability of frontal and posterior alpha power extracted from healthy and depressed adults.

Methods: Three, 8-minute resting EEG recordings were made for 89 adults (36 MDD) (4-minutes eyes-closed [EC] and eyes-open [EO]). Recordings 1 and 2 were separated by a day; recording 3 occurred about 90 minutes after recording 2, following acute stress. Automated processing used Artifact Subspace Reconstruction, Independent Components Analysis, and dipole fitting. Clean data were divided into 2 second epochs (75% overlap) and extracted from 24 electrodes: 12 frontal, 12 parietal (6 left/right hemisphere). Pearson correlations examined individual differences in frontal/parietal alpha power across sessions. Group x Condition x Hemisphere ANOVAs on alpha power, run separately for frontal and parietal scalp regions, characterized the data.

Results: Individual differences in frontal and parietal alpha were reliable, especially during EC (rs > 0.84) vs. EO (rs > 0.60), Z > 3.3. Alpha power was greater at parietal vs. frontal electrodes (Region effect, p < 0.001) and after stress (Session effect, p < 0.001). Power was also greater for EC vs. EO (Condition effect, ps < 0.001), and for the right vs. left hemisphere at parietal—but not frontal—electrodes (parietal Hemisphereeffect, ps < 0.001). No significant effects or interactions involving Group emerged.

Conclusions: Automated preprocessing was efficient and yielded expected effects of location (parietal > frontal) and condition (EC > EO) on alpha power, validating the approach. Individual differences in frontal and parietal alpha were similarly stable over time, and alpha increased post-stress. The lack of group differences was surprising; ongoing analyses are examining possible MDD effects in other frequency bands.