OBJECTIVES: To compare frequency analysis to human raters and determine the interrater agreement of postictal EEG changes after focal seizures.
METHODS: 24 focal seizures with and without impaired awareness recorded with scalp-EEG in the epilepsy monitoring unit were selected. Five board-certified neurophysiologists annotated seizure termination and end of postictal changes for all seizures. We assessed agreement using intraclass correlation, described the band-power changes by time-frequency analysis, and correlated these measures with the rater annotations.
RESULTS: Interrater agreement on the duration of the postictal changes was moderate (0.64, 95% confidence interval: 0.36-0.82). The interrater agreement for seizure termination was excellent (1.00). Median duration of the postictal interval of seizures with impaired awareness was significantly shorter than for seizures with retained awareness (p?=?0.0004). Mean postictal duration was 16.4?min. Seizure duration did not predict duration of the postictal changes. We found a strong correlation of 0.8 between the median human rater and the duration of the decrease in spectral edge frequency.
CONCLUSIONS: The agreement of neurophysiologists is moderate for duration of postictal changes and high for seizure termination. Rater determination of postictal duration is correlated with measures of EEG slowing.
SIGNIFICANCE: Disagreement between neurophysiologists on postictal duration need to be considered.