Complexity Analysis For Electroencephalogram During Anesthesia

Anesthesia during surgery can keep patients from pain and sufferings and make doctors work more smoothly. Anesthesia is patients' non-reaction and non-recollection to stimulation from hurt. It emphasizes to the restraint of consciousness and response to hurt stimulations. But inappropriate anesthesia cannot eliminate patients' pain and sufferings further more it will bring other problems. Over anesthesia may bring patients with nerve sequela and even death. While insufficient anesthesia may cause awareness that will result in serious nerve or sleep obstacles. Therefore a kind of valid method for anesthesia depth monitoring is in urgent clinical requirement to control anesthesia at appropriate level. However, now the anesthetists still judge depth of anesthesia (DOA) primarily depending on body characteristics such as breath, heart rate, blood pressure, body temperature, skeleton muscle responds and so on.During recent years the research to find an objective quantifier for anesthesia depth monitoring becomes a hotspot. Many researchers were dedicated to look for parameters that can represent DOA from physiological signals. Among them the researches on electroencephalogram (EEG) have showed best performance as general anesthesia block consciousness by depressing the central nervous system and EEGs reflect the electrical activity of the cerebral cortex. Nevertheless a valid index for DOA monitoring that will be accepted by anesthetists is still lacking. For over half a century, some processed EEG derivatives, like peak-to-peak amplitude, duration of each wave (time domain), spectral edge frequency, median frequency, frequency band-power ratio (frequency domain), have been developed. But none of these methods has been shown sufficient reliability for general use in clinical anesthesia, as all these methods are based on the assumption that the EEG is stationary. In reality, the EEGs exhibit strong feature of non-linearity and non-stationarity. Recently the development and application of complexity analysis have enlightened the work. Nonlinear complexity analyzing methods are moresuited for EEGs analysis and able to extract latent dynamic intormation from EEG signals.In this dissertation all nonlinear complexity analyzing methods and their applications in EEGs processing were summarized. We selected three complexity measures-Wavelet entropy (WE), Kc complexity (C(n)) and approximate entropy (ApEn) to analysis 31 cases EEGs recorded during anesthetic experiments. The results showed Wavelet entropy, Kc complexity and approximate entropy of EEGs obtained from awake state and asleep state respectively all exist obvious difference. The WE and Kc complexity of EEGs showed good performance in distinguishing awake state and asleep state by a threshold method, while ApEn quantifier failed. The WE application need less data and is computationally fast to fit clinical real-time use. Furthermore and the wavelet methods can make multiresolution analysis and eliminate noise at same time. WE analysis is a potential tool to analyze dynamic property of EEGs and it will enlighten the researches of DOA monitoring.