Assessment Of Cue-induced Craving For Methamphetamine Abusers Using Eeg Spectral Features

Objectives: Methamphetamine(MA)dependence is a major public health concern.Craving lies at the core of addiction.Traditionally,assessment of craving mostly relies on picture or video cues,which were less of ecological validity.Moreover,self-report measures may influence the objectiveness of results.Previous studies have shown that electroencephalogram(EEG)is an objective and precise method for detecting brain dysfunction in people with substance use disorder,and EEG abnormalities may reflect the underlying changes in brain function caused by chronic drug abuse.Thus,the aim of present study is to assess cue-induced craving for MA-dependent individuals in a virtual reality environment using EEG spectral features.Methods:(1)41 male methamphetamine addicts and 40 healthy males were enrolled and divided into two groups.EEG signals and subjective craving scores were recorded in resting state and in virtual reality cue-induced environment,respectively.The relative power of δ(1~4 Hz),θ(4~8 Hz),α(8~13 Hz),β(13~32 Hz),γ(32~49 Hz)bands were analyzed.A linear discriminant analysis(LDA)model was used to classify addicts and healthy controls;(2)29 male MA addicts and 30 healthy males were enrolled and divided into two groups.EEG signals and subjective craving scores were recorded in resting state and in virtual reality(VR)cue-induced environment,respectively.MA group then received a VR counterconditioning procedure(VRCP).After VRCP,EEG signals and subjective craving scores were recorded with the same settings as mentioned above.Spectrum power(d B)in the gamma range was analyzed for each condition and groups.Results:(1)For MA group,the relative power in δ,θ and α bands was significantly lower than healthy controls,and the relative power in β and γ bands was significantly higher than healthy controls(P < 0.05).Using a supervised machine learning algorithm with the power spectral features extracted from EEG activities,the discriminant model classified healthy controls and addicts with accuracies of 92.5%(37/40)and 90.2%(37/41),respectively.(2)Subjects with methamphetamine dependence showed significantly stronger self-reported craving and higher gamma power in VR environment than healthy individuals.VR environment elicited a significant increase in gamma power in MA abusers,compared with the resting state.After VRCP,participants showed significantly lower self-reported craving scores and gamma power when exposed to drug-related cues than the first time.Conclusion: The EEG power spectral features of MA group are significantly different from normal people,and could be used as biomarkers to detect MA craving.Drug-related VR environment elicited a greater increase of gamma power in patients with MA-dependence than healthy controls.EEG gamma band power may be a marker of cueinduced reactivity in patients with MA dependence.