[An anesthesia depth computing method study based on wavelet transform and artificial neural network].

Area of Science:

• Anesthesiology
• Biomedical Engineering
• Signal Processing

Background:

• General anesthesia is crucial for surgical patient safety.
• Electroencephalogram (EEG) monitoring reflects brain activity during anesthesia.
• Accurate assessment of anesthesia depth is vital for patient care.

Purpose of the Study:

• To develop and evaluate a novel method for assessing anesthesia depth.
• To combine wavelet transform and artificial neural networks (ANN) for EEG signal analysis.
• To differentiate between four levels of anesthesia: awake, light, moderate, and deep.

Main Methods:

• EEG signals were decomposed using discrete wavelet transform.
• Nine characteristic parameters were extracted from wavelet coefficients.
• Kruskal-Wallis test confirmed statistical significance of parameters (P < 0.001).
• ANN model utilized these parameters with Bispectral Index (BIS) as reference.

Main Results:

• The proposed method achieved 85.98% accuracy in classifying anesthesia levels.
• A high correlation coefficient of 0.977 was found with the BIS.
• The method demonstrated significant statistical differences across anesthesia states.

Conclusions:

• The combined wavelet transform and ANN method effectively distinguishes anesthesia depth.
• This approach shows promise for real-time anesthesia depth monitoring during surgery.
• Further application prospects exist for improving patient safety in general anesthesia.