Bicoherence of EEG can be used to differentiate anesthetic levels

Incidents of intra-operative awareness can result in severe, long-lasting post-traumatic distress syndrome. Fine tuning “depth of anesthesia” is clearly a desirable objective to prevent underdosing and overdosing. Various EEG derived variables have been used toward this objective with some success, and recent advances in computer technology have made bispectral analysis feasible for implementation in a real-time clinical monitor. Bispectral analysis is a sophisticated tool (capable of detecting inter-frequency phase-coherence) which the authors have applied to characterization of the EEG during anesthesia. In this study artificial neural networks (ANN) were used with bispectral analysis patterns for classification of EEG during propofol anesthesia with intentional “awareness incidents.”