The aim is to construct a method, based upon statistical pattern recognition techniques, including neural networks, whereby awareness during general anaesthesia may be detected. The data source for this system would be a single channel of the electroencephalogram (EEG). Pre-processing of data prior to input into the network is a critical component of the work, and it is here that parametric models have been utilised. A spectral representation has been extracted from the EEG based upon 1 second of data, using a lattice filter as the primary model; and a bispectral representation based upon 5 seconds of data has also been constructed, this time using a transversal filter as the underlying model.
展开▼