頁籤選單縮合
| 題 名 | Neural Network Based Prediction of Depth of Anaesthesia Using Auditory Evoked Potentials with a Wavelet Transform |
|---|---|
| 作 者 | Backory,J. K.; Linkens,D. A.; Peacock,J. E.; | 書刊名 | 醫學工程 |
| 卷 期 | 10:4 1998.08[民87.08] |
| 頁 次 | 頁41-48 |
| 專 輯 | Special Issue on Control Methods in Anaesthesia |
| 分類號 | 410.1644 |
| 關鍵詞 | Anaesthesia; Wavelets; Neural networks; Auditory evoked responses; Feature extraction; |
| 語 文 | 英文(English) |
| 英文摘要 | We investigate the reliability of predicting depth of anaesthesia using features extracted from the Middle Latence Auditory Evoked Potentials using a wavelet transform by classifying the features using neural networks. Fourteen patients undergoing surgery were anaesthetised using Total Intraverous Anaesthesia with propofol, except for one who was anaesthetised using enflurane. The brain signals were revorded continuously durign the period of general anaesthesia and these were later averaged off-line to generate the Middle Latency Auditory Evoked Potentials. A Multiresolution Wavelet Analysis of these signals was then carried out and features were obtained from a selection of the wavelet coefficients. Data from nine patients were used to train and test a Kohonen self-organising feature map, a Quickprop-trained multilayer perceptron, and a radial basis function respectively, and the data from the other five patients were used to validate the classifiers. The performance of the classifiers was found to be acceptable given the complexity of the problem and the problems associated with assessing anaesthetic depth. The radial basis function correctly classified 68 % of the validation data with most of the misclassifications occurring, as expected, in the OK and OK Light planes of anaesthesia. |
本系統中英文摘要資訊取自各篇刊載內容。