Recognizing of stereotypic patterns in epileptic EEG using empirical modes and wavelets

• Published in: Physica A Vol. 486; pp. 206 - 217
• Main Authors: Grubov, V.V., Sitnikova, E., Pavlov, A.N., Koronovskii, A.A., Hramov, A.E.
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.11.2017
• Subjects: Electroencephalogram; Empirical mode decomposition; Patterns; Wavelet-analysis; Empirical mode decomposition; Wavelet-analysis; Electroencephalogram; Patterns
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2017.05.091

Epileptic activity in the form of spike–wave discharges (SWD) appears in the electroencephalogram (EEG) during absence seizures. This paper evaluates two approaches for detecting stereotypic rhythmic activities in EEG, i.e., the continuous wavelet transform (CWT) and the empirical mode decomposition (EMD). The CWT is a well-known method of time–frequency analysis of EEG, whereas EMD is a relatively novel approach for extracting signal’s waveforms. A new method for pattern recognition based on combination of CWT and EMD is proposed. It was found that this combined approach resulted to the sensitivity of 86.5% and specificity of 92.9% for sleep spindles and 97.6% and 93.2% for SWD, correspondingly. Considering strong within- and between-subjects variability of sleep spindles, the obtained efficiency in their detection was high in comparison with other methods based on CWT. It is concluded that the combination of a wavelet-based approach and empirical modes increases the quality of automatic detection of stereotypic patterns in rat’s EEG. •An automatic method for detection of stereotypic patterns in EEG is proposed.•The method is based on combination of wavelets and empirical mode decomposition.•Such combination increases the quality of patterns detection in rodents EEG.•The detection of sleep spindles can be done with high sensitivity and specificity.