Method for the Automatic Detection of Epileptiform Waveforms in Sevoflurane-induced Anesthesia EEG

Abstract

Sevoflurane is a volatile anesthetic which is reported to cause epileptiform EEG changes together with undesired symptoms such as convulsions. In this paper, an algorithm for the automatic detection of these EEG changes is presented which could enable safer induction of anesthesia with sevoflurane by informing the clinicians about the epileptiform EEG. EEG was recorded from 60 healthy female patients during sevoflurane anesthesia. A neurophysiologist classified the EEG waveforms. Each anesthesia period lasted 6 minutes. 48 signal features were extracted from the raw EEG. 5-sec segments of EEG were classified based on the extracted features using a decision tree with a logistic regression based decisions and the classification results were compared to the neurophysiologist's classifications. Awake EEG was recognized with 69% / 96% (sensitivity / specificity), Burst suppression with 56% / 98%, Epileptiform EEG with 83% / 87 %, normal slow anesthesia EEG with 86% / 64 %, slow anesthesia EEG with monophasic pattern with 65% / 80 %, and slow anesthesia EEG with monophasic pattern and spikes with 54% / 84%.