Monitoring Depth of Anesthesia Using Detrended Fluctuation Analysis Based on EEG Signals
Detrended fluctuation analysis (DFA) is appropriate for the analysis of long-range correlation in nonstationary time series. In this study, DFA was used to study electroencephalography (EEG) fluctuations in order to assess the depth of anesthesia (DOA) and measure the level of consciousness. The fluctuation function F(s) was calculated. The distribution of F(s) in segments was used to classify anesthesia state levels into awake, light, moderate, and deep states. A linear fit of F(s) versus s in each segment was performed. Finally, the point at which the fitted line crossed the defined line divided the four zones corresponding to the four states of DOA from 100 to 0 in the coordinate system. Experimental results demonstrate that the proposed method can accurately identify the states of DOA based on EEG signals. The ranges of DOA values can be extended through adjustable parameters, improving the adaptability of the algorithm. The results are close to the bispectral index values, which can be used to identify anesthesia states. The proposed DFA method is effective for monitoring DOA.