Depth of anesthesia assessment and higher brain function modelling for consciousness

Anaesthesia is the corner stone of modern surgical medicine. Despite a long period of enquire beginning with Snow (1847) anaesthesia remains a field in which there are more questions than answers. This thesis reports findings on three different aspects of anaesthesia. 1. Initially, a method for calculating a population pharmacokinetic model for propofol infusion is described. This method greatly reduced the time required to calculate the model (0.1 seconds per iteration) compared to the NONMEM method (hours per iteration (Minto, Schnider, Egan, Youngs, Lemmens, Gambus,Billard, Hoke, Moore, Hermann, Muir, Mandema & Shafer 1997)). The resultant model achieved improved fit to the data than the model of Schuttler & Ihmsen (2000b) achieving a mean squared error of 0.2835 compared to 0.6413 respectively. 2. Second, a neural network (NN) method is presented to assess Depth of Anaesthesia from long segments of raw EEG. The proposed method was able to approximate the output from a BIS XP monitor for the training data. The linear regression, between the NN and the BIS monitor, resulted in an R value of 0.99963. The network was able to approximate the BIS monitor output for new (unseen) data. 3. Finally, a lumped parameter neural mass, anaesthesia, model is presented. This model is capable of generating changes in EEG associated with increasing doses of Gamma-aminobutyric acid type A (GABA A) hypnotic agent (propofol). This model was not a fitting exercise rather it was constructed based on known brain physiology, and the changes to Alpha1 GABA A receptors conductance caused by propofol. Encompassing the regional interactions, that are thought to be, altered by GABA hypnotic agents. The model is capable of producing five distinct EEG patterns (Beta, Alpha, Theta, Delta and isoelectric) in response to different levels of hypnotic agent. The model is reactive capable of switching from Alpha to Beta band EEG when the eyes open. Anaesthetic supresses the models transition to a higher state EEG. The model suggest that the effect site for propofol as Alpha1 GABA A receptors of slow interneurons of the cortex.