Automatic detection and analysis of the EEG sharp wave–slow wave patterns evoked by fluorinated inhalation anesthetics

ObjectiveThe aim of this study was to develop a method for the automatic detection of sharp wave–slow wave (SWSW) patterns evoked in EEG by volatile anesthetics and to identify the patterns’ characteristics.MethodsThe proposed method consisted in the k-NN classification with a reference set obtained using expert knowledge, the morphology of the EEG patterns and the condition for their synchronization. The decision rules were constructed and evaluated using 24 h EEG records in ten patients.ResultsThe sensitivity, specificity and selectivity of the method were 0.88 ± 0.10, 0.81 ± 0.13 and 0.42 ± 0.16, respectively. SWSW patterns’ recruitment was strictly dependent on anesthetic concentration. SWSW patterns evoked by different types of anesthetics expressed different characteristics.ConclusionsSynchronization criterion and adequately selected morphological features of “slow wave” were sufficient to achieve the high sensitivity and specificity of the method.SignificanceThe monitoring of SWSW patterns is important in view of possible side effects of volatile anesthetics. The analysis of SWSW patterns’ recruitment and morphology could be helpful in the diagnosis of the anesthesia effects on the CNS.

► Epileptiform patterns that possess both “sharp wave” and “slow wave” components (SWSW patterns) appear during volatile anesthesia in anesthetic concentration- and anesthetic type-dependent way. ► The novelty of the proposed method of automatic detection of epileptiform patterns during anesthesia consists in the analysis of morphology of individual SWSW patterns and not, as it was the case so far, on the quantification of 5s EEG segments containing monophasic activity and/or spikes. ► The on-line analysis of SWSW patterns’ rate and morphology offered by the proposed method is important in view of possible side effects of volatile anesthetics.