Predicting depth of anesthesia using bispectral parameters in neural networks

Features like the spectral edge and median frequency derived from power spectrum of the EEG have so far failed to show any consistent changes with the depth of anesthesia. One of the disadvantages of using power spectrum is that it suppresses phase information in the signal. A third order spectrum or bispectrum preserves phase information. A bispectral parameter called bicoherence index was derived from the EEG prior to a tail clamp. Using the bicoherence index and the estimated MAC level of the dog at that time a neural network was able to correctly classify all the 36 data points from a test group corresponding to either an awake or an asleep dog