Brain and brain-heart Granger causality during wakefulness and sleep

Abstract

In this exploratory study we apply Granger Causality (GC) to investigate the brain-brain and brain-heart interactions during wakefulness and sleep. Our analysis includes electroencephalogram (EEG) and electrocardiogram (ECG) data during all-night polysomnographic recordings from volunteers with apnea, available from the Massachusetts General Hospital's Computational Clinical Neurophysiology Laboratory and the Clinical Data Animation Laboratory. The data is manually annotated by clinical staff at the MGH in 30 second contiguous intervals (wakefulness and sleep stages 1, 2, 3, and rapid eye movement (REM). We applied GC to 4-s non-overlapping segments of available EEG and ECG across all-night recordings of 50 randomly chosen patients. To identify differences in GC between the different sleep stages, the GC for each sleep stage was subtracted from the GC during wakefulness. Positive (negative) differences indicated that GC was greater (lower) during wakefulness compared to the specific sleep stage. The application of GC to study brain-brain and brain-heart bidirectional connections during wakefulness and sleep confirmed the importance of fronto-posterior connectivity during these two states, but has also revealed differences in ipsilateral and contralateral mechanisms of these connections. It has also confirmed the existence of bidirectional brain-heart connections that are more prominent in the direction from brain to heart. Our exploratory study has shown that GC can be successfully applied to sleep data analysis and captures the varying physiological mechanisms that are related to wakefulness and different sleep stages.

Keywords: Granger causality; connectivity; electrocardiogram (ECG); electroencephalogram (EEG); sleep.

FIGURE 1

Topographies of significant GC differences between wakefulness and sleep [(A) Stage 1, (B) stage 2, (C) stage 3, (D) REM], at F3, F4, C3, C4, O1, and O2 electrode locations. Results are averaged over all 50 participants. Dashed lines indicate negative differences, i.e., GC during the specific sleep stage is significantly greater than GC during wakefulness. Solid lines indicate positive differences, i.e., GC during the specific sleep stage is significantly lower than GC during wakefulness. The line thickness corresponds to the strength of the difference, i.e., thicker lines represent larger differences compared to thinner lines.

FIGURE 2

Fronto-posterior significant average GC differences between wakefulness and sleep stages S1, S2, S3, and REM. Negative values indicate GC is greater during sleep compared to wakefulness, and vice versa. In the posterior→frontal direction only the contralateral GC was significant, while in the fronto→posterior direction both contralateral and ipsilateral connectivity were significant.

FIGURE 3

Significant average GC between brain-heart at different EEG locations and across wakefulness and sleep. GC is significant in both directions, but the dominant direction of interaction is brain→heart, with the exception of C3.