Depth of Anesthesia using Neural Networks

Abstract The need for a reliable method of measuring the depth of anesthesia has existed since the introduction of anesthesia. Electroencephalograms (EEG), a record of brain activity, can be used to monitor the anesthetic depth. This paper proposes a recognition system, based on the autoregressive (AR) modeling and neural network analysis of the EEG signals, to monitor the depth of anesthesia. The input to the neural network will be the AR parameters along with the hemodynamic parameters blood pressure (BP) and heart rate (HR), and the output will be the depth of anesthesia. Design of the system and results from the preliminary tests on dogs are presented in this paper. The experiments were carried out on ten dogs at different levels of halothane. Depth of anesthesia was tested by monitoring the response to tail clamping, which is considered to be a supra-maximal stimulus in dogs. The EEG data obtained prior to tail clamping was processed using a tenth order AR model and the parameters obtained were used as input to a three layer perceptron feedforward neural network. The network was able to correctly classify the depth in 85% of the cases as compared to 65% when only hemodynamic parameters were used as the input to the network. When both the measures were combined, the recognition rate rose to greater than 92%. This recognition system shows the feasibility of using the EEG signals for monitoring the depth of anesthesia.

[1]  Frederick R. Fromm,et al.  Class: A nonparametric clustering algorithm , 1976, Pattern Recognit..

[2]  R. Roy,et al.  An adaptive controller for the administration of closed-circuit anesthesia during spontaneous and assisted ventilation , 1993, Journal of clinical monitoring.

[3]  Ben H. Jansen,et al.  Autoregressive Estimation of Short Segment Spectra for Computerized EEG Analysis , 1981, IEEE Transactions on Biomedical Engineering.

[4]  Richard P. Lippmann,et al.  An introduction to computing with neural nets , 1987 .

[5]  Stanley C. Fralick,et al.  Learning to recognize patterns without a teacher , 1967, IEEE Trans. Inf. Theory.

[6]  G. B. Anderson,et al.  Monitoring the Level of Anesthesia by Automatic Analysis of Spontaneous EEG Activity , 1975, IEEE Transactions on Biomedical Engineering.

[7]  J. P. Burg,et al.  Maximum entropy spectral analysis. , 1967 .

[8]  Andrew K. C. Wong,et al.  PFS Clustering Method , 1979, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  H. Shapiro,et al.  Automated EEG processing for intraoperative monitoring: a comparison of techniques. , 1980, Anesthesiology.

[10]  H. Akaike.  Fitting autoregressive models for prediction , 1969 .

[11]  D R Stanski,et al.  Pharmacodynamic modeling of anesthetic EEG drug effects. , 1992, Annual review of pharmacology and toxicology.

[12]  W W Mapleson,et al.  Circulation-time models of the uptake of inhaled anaesthetics and data for quantifying them. , 1973, British journal of anaesthesia.

[13]  E. Eger,et al.  Minimum alveolar anesthetic concentration: a standard of anesthetic potency. , 1965, Anesthesiology.

[14]  G. Mirchandani,et al.  On hidden nodes for neural nets , 1989 .

[15]  Carsten Thomsen,et al.  COMPUTERIZED MONITORING OF DEPTH OF ANAESTHESIA WITH ISOFLURANE , 1989 .

[16]  W W Mapleson,et al.  The solubility of halothane in canine blood and tissues. , 1975, British journal of anaesthesia.

[17]  Lewis B. Sheiner,et al.  Simultaneous modeling of pharmacokinetics and pharmacodynamics: an improved algorithm , 1987, Comput. Appl. Biosci..

[18]  T. Ulrich,et al.  Maximum entropy spectral analy-sis and autoregressive decomposition , 1975 .

[19]  Rob J. Roy,et al.  Adaptive control of closed-circuit anesthesia , 1991 .

[20]  Bernard Widrow,et al.  30 years of adaptive neural networks: perceptron, Madaline, and backpropagation , 1990, Proc. IEEE.

[21]  Kishan G. Mehrotra,et al.  Bounds on the number of samples needed for neural learning , 1991, IEEE Trans. Neural Networks.