Epileptic seizure onset detection based on EEG and ECG data fusion

Epilepsy Behav. 2016 May;58:48-60. doi: 10.1016/j.yebeh.2016.02.039. Epub 2016 Apr 5.


This paper presents a novel method for seizure onset detection using fused information extracted from multichannel electroencephalogram (EEG) and single-channel electrocardiogram (ECG). In existing seizure detectors, the analysis of the nonlinear and nonstationary ECG signal is limited to the time-domain or frequency-domain. In this work, heart rate variability (HRV) extracted from ECG is analyzed using a Matching-Pursuit (MP) and Wigner-Ville Distribution (WVD) algorithm in order to effectively extract meaningful HRV features representative of seizure and nonseizure states. The EEG analysis relies on a common spatial pattern (CSP) based feature enhancement stage that enables better discrimination between seizure and nonseizure features. The EEG-based detector uses logical operators to pool SVM seizure onset detections made independently across different EEG spectral bands. Two fusion systems are adopted. In the first system, EEG-based and ECG-based decisions are directly fused to obtain a final decision. The second fusion system adopts an override option that allows for the EEG-based decision to override the fusion-based decision in the event that the detector observes a string of EEG-based seizure decisions. The proposed detectors exhibit an improved performance, with respect to sensitivity and detection latency, compared with the state-of-the-art detectors. Experimental results demonstrate that the second detector achieves a sensitivity of 100%, detection latency of 2.6s, and a specificity of 99.91% for the MAJ fusion case.

Keywords: ECG/EEG fusion; Heart rate variability; Seizure onset detection; Time–frequency analysis.

MeSH terms

  • Adult
  • Aged
  • Algorithms
  • Brain / physiopathology*
  • Electrocardiography / methods*
  • Electroencephalography / methods*
  • Female
  • Heart Rate / physiology*
  • Humans
  • Male
  • Middle Aged
  • Seizures / diagnosis*
  • Seizures / physiopathology
  • Sensitivity and Specificity
  • Signal Processing, Computer-Assisted