A Kalman-based instantaneous frequency estimation for anesthetic depth measurement

Fast and accurate measuring depth of anesthesia (DoA) during heavy surgeries (e.g. orthopedic or neurosurgery) is still a challenge. Late estimation of DoA in critical conditions may lead to severe effects such as a comma or conscious state, and jeopardize patient's life accordingly. Recently, several attempts have been made to elicit an accurate DoA index by analyzing electroencephalogram (EEG) signals, where the pain is reflected. Previously, it is shown that elicited instantaneous frequency (IF) of EEG signals of patients has a noticeable correlation with the corresponding gold-standard bispectral index scale (BIS) index. In this study, the relevance of IF is improved by applying a preprocessing stage in which a recursive band-pass filter is applied to raw EEGs for selecting the anesthetic-dependent band of EEG. To do this, Kalman filter is employed to update cutoff frequencies of the band-pass filter through successive windows. Empirical results over the four anesthetized patients imply on the superiority of the proposed scheme, compared to its counterparts, in terms of speed and accuracy.