Extraction of anesthesia depth using self similarity and fluctuation analysis on the wavelet coefficients of EEG

The depth of anesthesia estimation has been of great interest in recent decades. In this paper, we present a new methodology to quantify the levels of consciousness. Our algorithm takes advantage of the fractal and self-similarity properties of the EEG signal. We have tried to find the effect of anesthetic agents by using the detrended fluctuation analysis (DFA) as a self similarity estimator of a fractal process. The implementation results confirm that the DFA on the raw EEG data can clearly discriminate between aware to moderate anesthesia levels, but the moderate to deep anesthesia cannot be discriminated. We have extended the idea by considering that the self-similarity property of fractal signal has a better resolution on the wavelet domain. By applying the DFA on different scales of wavelet coefficients and quantifying the relative drift between the lines generated by DFA, the depth of anesthesia can be discriminated precisely.