Light general anesthesia during surgery has the advantages of producing less depression of physiologic functions and allowing the patient to awaken more rapidly afterwards. Sometimes, however, a patient under light anesthesia arouses enough to move in response to surgical stimulation or even to recall later what happened. Supplementing traditional monitoring with a monitor of the background electroencephalogram (EEG) could help the anesthesiologist titrate the administration of anesthetic agents to avoid incidents of arousal, especially if the EEG could be reduced to a scalar index of patient state.
The specific aims of this study are to use existing data to identify correlations between characteristics of the EEG frequency spectrum and the patient's level of arousal, to use these correlations to develop simple scalar indexes of level of arousal, and to determine how well the indexes can predict level of arousal during surgery. EEG frequency spectra have been recorded on magnetic diskette during surgery on 150 patients, along with blood pressure, heart rate, and delivered anesthetic and event data. Anesthesia was maintained using one of four techniques: fentanyl with isoflurane, isoflurane with N2O, fentanyl with N2O, or isoflurane. The spectra were computed on-line using an autoregressive implementation of Spectral Component Parameter Analysis (SPA). This method of power spectral analysis rapidly and reliably produces spectral estimates in a concise form that is very convenient for both visual interpretation and further mathematical processing. The level of arousal was classified on the basis of blood pressure and heart rate and whether there was movement in response to surgical stimulation, movement to verbal requests, or statistical evidence of memory. Visual analysis of the spectra suggests strong correlations between spectral pattern and level of arousal for the given anesthetic techniques. This study will apply Discriminant Analysis to training sets of spectral data to develop scalar indexes of level of arousal. Performance in predicting level of arousal from test set data will then be used to determine the best derived indexes and to compare them to existing scalar EEG indexes (e.g., average power, peak frequency, median frequency, spectral edge).
| Dutton, R C; Smith, W D; Smith, N T (1996) EEG Predicts movement response to surgical stimuli during general anesthesia with combinations of isoflurane, 70% N2O, and fentanyl. J Clin Monit 12:127-39 |
| Smith, W D; Dutton, R C; Smith, N T (1996) Measuring the performance of anesthetic depth indicators. Anesthesiology 84:38-51 |
