Semi-automatic quantification of seizure-related effects on heart activity

Seizure-related modulation of heart rate (HR) was examined extensively in previous studies. However, the overall effect on HR attributable to epileptic seizures is difficult to determine, given the considerable fluctuations of HR before and during seizures. Here, we developed a semi-automatic procedure allowing quantification of the total impact of seizures on HR and determination of temporal relationships between seizure onset assessed by intracranial EEG (iEEG) and ECG.

ECG and iEEG data of epilepsy patients undergoing video-EEG telemetry for epilepsy surgery with bilateral hippocampal depth electrodes were analysed retrospectively. Consecutive RR intervals and HR profiles were determined using R detection algorithms. Novel features including the normalized ictal area under the curve (niAUC), as well as the time point of ECG onset (HR breakpoint) were calculated. Selected HR features were compared to widely-used manually acquired measures. Data are given as median ± SD.

Fifteen patients had a total of 34 seizures with left-hippocampal and 37 seizures with right-hippocampal onset. HR increased by 9 ± 19% during seizures. Latency between iEEG seizure-onset to the HR breakpoint was 23 ± 22 s. No significant difference between left- and right-hippocampal seizures was observed with respect to HR increases, latencies and niAUC. A comparison between results of the semi-automatic and manual approach revealed that ictal HR changes showed a higher correlation (r = 0.6) than niAUC (r = 0.4).

The proposed semi-automatic approach to analyze continuous HR data provides useful tools for estimating the overall effect of seizures on HR in greater detail. Our results suggest that the side of hippocampal seizure onset has no significant effect on the latency and extent of ictal HR changes. The algorithms may be of further use in clinical research and the development of seizure detection devices.