Correlations between the signal complexity of cerebral and cardiac electrical activity: a multiscale entropy analysis

Abstract

The heart begins to beat before the brain is formed. Whether conventional hierarchical central commands sent by the brain to the heart alone explain all the interplay between these two organs should be reconsidered. Here, we demonstrate correlations between the signal complexity of brain and cardiac activity. Eighty-seven geriatric outpatients with healthy hearts and varied cognitive abilities each provided a 24-hour electrocardiography (ECG) and a 19-channel eye-closed routine electroencephalography (EEG). Multiscale entropy (MSE) analysis was applied to three epochs (resting-awake state, photic stimulation of fast frequencies (fast-PS), and photic stimulation of slow frequencies (slow-PS)) of EEG in the 1-58 Hz frequency range, and three RR interval (RRI) time series (awake-state, sleep and that concomitant with the EEG) for each subject. The low-to-high frequency power (LF/HF) ratio of RRI was calculated to represent sympatho-vagal balance. With statistics after Bonferroni corrections, we found that: (a) the summed MSE value on coarse scales of the awake RRI (scales 11-20, RRI-MSE-coarse) were inversely correlated with the summed MSE value on coarse scales of the resting-awake EEG (scales 6-20, EEG-MSE-coarse) at Fp2, C4, T6 and T4; (b) the awake RRI-MSE-coarse was inversely correlated with the fast-PS EEG-MSE-coarse at O1, O2 and C4; (c) the sleep RRI-MSE-coarse was inversely correlated with the slow-PS EEG-MSE-coarse at Fp2; (d) the RRI-MSE-coarse and LF/HF ratio of the awake RRI were correlated positively to each other; (e) the EEG-MSE-coarse at F8 was proportional to the cognitive test score; (f) the results conform to the cholinergic hypothesis which states that cognitive impairment causes reduction in vagal cardiac modulation; (g) fast-PS significantly lowered the EEG-MSE-coarse globally. Whether these heart-brain correlations could be fully explained by the central autonomic network is unknown and needs further exploration.

Figure 1. Electrocardiography (ECG) and electroencephalography (EEG).

One cycle of ECG includes various deflections, P, Q, R, S (QRS complex) and T. All R peaks of ECG recordings were detected to obtain the RR interval (RRI) time series. Each EEG recording includes brain waves from 19 electrode sites, one ECG recording and one trace of photic stimulation (PS).

Figure 2. Examples of RR interval (RRI), detrended EEG and the multiscale entropy (MSE) profiles of them.

One RRI time series and one detrended single-channel EEG signal are shown on the left-hand side of the figure. MSE-RRI 1-3 are examples of the MSE profiles showing the MSE values of RRI from scale 1 to scale 20, while MSE-EEG 1-3 are examples of the MSE profiles showing the MSE values of EEG from scale 1 to scale 20. All the MSE-profiles show an initial increasing before a plateau or a fall.

Figure 3. Inverse association between the multiscale entropy (MSE) values of the awake RRI and resting-awake EEG.

(a) Regions with significant inverse correlation between the summed MSE values on the scales 11−20 of the awake RRI and the summed MSE values on the scales 6−20 of the resting-awake EEG after Bonferroni corrections (corrected p-values  =  original p-values19, alpha  =  0.05). r and p denote the Pearson’s partial correlation coefficient and corrected significance level, respectively. (b) The brain map illustrates regions with significant association. The relative brightness is according to the sequential p-values from the smallest one (Fp2, C4, T6 and T4).

Figure 4. Inverse association between the multiscale entropy (MSE) values of the awake RRI and fast-PS EEG.

(a) Regions with significant inverse correlation between the summed MSE values on the scales 11−20 of the awake RRI and the summed MSE values on the scales 6−20 of the photic-simulated EEG at frequency 12, 15, 18 and 24 Hz (fast-PS, duration 10 seconds and interval 10 seconds) after Bonferroni corrections (corrected p-values  =  original p-values19, alpha  =  0.05). r and p denote the Pearson’s partial correlation coefficient and corrected significance level, respectively. (b) The brain map illustrates regions with significant association. The relative brightness is according to the sequential p-values from the smallest one (O1, O2 and C4).