Permutation Fuzzy Entropy And Its Application In The Analysis Of Eeg

EEG is a commonly used method of brain detection.In recent years,a large number of studies have shown that the human brain is a highly complex nonlinear system,EEG signal,as a record of brain activity,also showed a high degree of complexity and non-linear characteristics.Entropy,as an important nonlinear dynamic feature,can describe the disorder and chaos of the state of the system.It has been widely used in quantitative analysis of different states of the brain,such as diseases,cognitive tasks,and so on.Among them,the fuzzy entropy,as an improved index of approximate entropy and sample entropy,it has been recognized by the researchers to a certain extent,but the sensitivity of the index to noise is susceptible to the size of the index parameters.However,there are more noise interference in the process of EEG signal acquisition,so that the signal-tonoise ratio of the finally obtained signal is very low,so we proposed a index which called permutation fuzzy entropy in this paper.This index improved the fuzzy entropy by introducing the idea of sorting symbolization on the basis of fuzzy entropy.Then,through the comparative study,it is found that the anti-noise performance of the permutation fuzzy entropy is superior to the traditional index,permutation entropy and fuzzy entropy.Finally,the three entropy indexs were applied to the autodetection of epileptic EEG and the analysis of schizophrenia event related brain potential data.Specific work is as follows:Firstly,this paper aimed at the problem that fuzzy entropy is sensitive to the noise,put forward an improved index called permutation fuzzy entropy,which is based on fuzzy entropy;then the anti-noise ability of permutation fuzzy entropy is compared with fuzzy entropy and permutation entropy in the simulated EEG data combined with gaussian white noise.The simulation experimental results show that the antinoise ability of permutation fuzzy entropy is better than that of permutation entropy and fuzzy entropy,and its ability to resist noise will not worse than permutation entropy and fuzzy entropy by different settings of index parameters.Secondly,the permutation fuzzy entropy,fuzzy entropy and permutation entropy were used for the same epilepsy automatic detection framework,and the ability of three entropy detecting epilepsy were compared.The results of epilepsy automatic detection experiment showed that the permutation fuzzy entropy was more suitable for epilepsy detection than fuzzy entropy and permutation entropy.Thirdly,the event related brain potentials data of schizophrenia under conditions-test stimulus mode with lower signal-to-noise ratio were analyzed by using permutation fuzzy entropy,fuzzy entropy and permutation entropy,respectively.The experimental results show that the permutation fuzzy entropy is more effective than fuzzy entropy and permutation entropy to analyze sensory gated P50 defects in schizophrenic patients.The results of three experiments in this paper consistently show that,the antinoise ability of proposed permutation fuzzy entropy is better than the fuzzy entropy,at the same time also is better than that of permutation entropy,which is famous for its resistance to noise;and compared with the fuzzy entropy and permutation entropy,permutation fuzzy entropy is more suitable for analysis of EEG signal.