Study of significance of spectral and wavelet energy measures to detect the electrical onset of seizure

Abstract

The objective of this study is to assess the utility of spectral and wavelet energy measures in detecting electrical onset of seizure in patients with temporal lobe epilepsy (TLE). The scalp-recorded EEG data of 20 seizures from 11 TLE patients is used for this study. The spectral and wavelet energy in same set of frequency bands are calculated for each 4 s windowed EEG signal. Among the 14 measures, 3–6 Hz and 6–12 Hz band spectral and wavelet energy increases at electrical onset in 60% and 90% of the recorded seizures respectively. The spectral and wavelet energy in 1–3 Hz band increases in 40% of the recorded seizures. This study identifies the correlation between spectral and wavelet energy in same frequency bands. Hence the simple and efficient spectral energy measures are selected as significant features for the design of automated seizure detection system.