Unveiling Frequency-Specific Microstate Correlates of Anxiety and Depression Symptoms

Electroencephalography (EEG) microstates are canonical voltage topographies that reflect the temporal dynamics of resting-state brain networks on a millisecond time scale. Changes in microstate parameters have been described in patients with psychiatric disorders, indicating their potential as clinical biomarkers with broadband EEG signals (e.g., 1-30 Hz). Considering the distinct information provided by specific frequency bands, we hypothesized that microstates in decomposed frequency band could provide a more detailed depiction of the underlying psychological mechanism. In this study, with a large open access resting-state dataset (n = 203), we examined the properties of frequency-specific microstates and their relationship with emotional disorders. We conducted clustering on EEG topographies in decomposed frequency band (delta, theta, alpha and beta), and determined the number of clusters with a meta-criterion. Microstate parameters, including global explained variance (GEV), duration, coverage, occurrence and transition probability, were calculated for eyes-open and eyes-closed states, respectively. Their predictive power for the scores of depression and anxiety symptoms were identified by correlation and regression analysis. Distinct microstate patterns were observed across different frequency bands. Microstate parameters in the alpha band held the best predictive power for emotional symptoms. Microstates B (GEV, coverage) and parieto-central maximum microstate C (coverage, occurrence, transitions from B to C) in the alpha band exhibited significant correlations with depression and anxiety, respectively. Microstate parameters of the alpha band achieved predictive R-square of 0.100 for anxiety scores, which is much higher than those of broadband (R-square = -0.026, p < .01). These results suggested the value of frequency-specific microstates in predicting emotional symptoms.