Feature Analysis And Classification Of Resting EEG For Heroin Addicts

Heroin addiction is a social problem.It will affect human health to be serious to cause death.And it can also lead to family breakdown and affect social stability.In China,the problem caused by heroin addiction is particularly serious.As of 2015,the total number of heroin addicts reached 980 thousand,accounting for 41.8% of the total number of drug addicts.The treatment of heroin addiction includes early diagnosis,clinical detoxification and rehabilitation.However,for the diagnosis of heroin addiction and assessment of rehabilitation after detoxification,which mainly base on the history of heroin addicts and doctor’s experience which is lack of objective evaluation method index.Therefore,it is very important to find an objective evaluation method for the diagnosis and assessment of rehabilitation.In this paper,the main work is based on resting state EEG features to find the differences between heroin addicts and normal people,and explore an objective method of diagnosis of heroin addiction or assessment of rehabilitation after detoxification.Three groups resting state EEG data were collected.The first group was collected from 42 heroin addicts,the second group was collected from 27 normal people.And after five months,we collected 31 raw datasets from heroin addicts which form the third group.We calculated correlation dimension,Kolmogorov entropy,Lempel-Ziv complexity and power spectral characteristics of Delta,Theta,Alpha,Beta and Gamma rhythm,and compared the differences between heroin addicts and normal people.The results show that the absolute power of EEG signals in heroin addicts is lower than normal people.Delta,Beta and Gamma rhythms relative power are higher than the normal,the relative power of Alpha is lower than that of normal.The complexity of Theta and Gamma rhythm in heroin addiction group is lower than that in control group,and the complexity of Alpha rhythm is higher than that of control group.The correlation dimension,Kolmogorov entropy and Lempel-Ziv complexity consistently reflect the complexity of the Alpha rhythm in the frontal and central areas are significantly higher than the control group,which indicates that the nonlinear characteristics of the Alpha rhythm in frontal and central areas are important to study the injury of brain function of heroin addicts.The analysis of the influence of withdrawal time found that EEG Alpha rhythm of heroin addicts is significantly improved with the growth of withdrawal time.Finally,three kinds of classifiers are used to evaluate the ability of different features to discriminate heroin addicts from control subjects,and we obtained 74.31% correct rate from correlation dimension.In this paper,through the analysis of EEG signals,we found differences between heroin addicts and normal people from complexity of EEG signal and EEG spectrum,which indicates that EEG signal can be used in the accessory diagnosis of heroin addiction.And Alpha rhythm of heroin addicts was significantly improved with the growth of withdrawal time,which indicates EEG signal can be used in the accessory assessment of rehabilitation after detoxification.The findings of this study provide a new way for the application of EEG analysis in the diagnosis of heroin addiction and the evaluation of rehabilitation effect.