Analysis Of Eeg Signals Based On Music Therapy

In today's society,pressure is everywhere,and the mental sub-health of young and middle-aged people is widespread.For the people with weak psychological endurance,they are troubled by negative emotions for a long time,and are prone to develop into anxiety,hyperactivity,depression and other neurological psychological diseases.For the elderly,long-term emotional depression increases the incidence of neurodegenerative diseases(for example,Alzheimer's disease).In this paper,music therapy,electroencephalogram(EEG)and emotion analysis were combined to study the EEG signals and emotion classification of left and right brain regions under different music environments.After analysis,it can be concluded that: under the positive music stimulation,the sensitivity of the left brain area is better and there is emotional laterality;under the negative music stimulation,the sensitivity of the right brain area is better and there is emotional laterality.In addition,it is found that the average emotion classification accuracy of EEG signals in the left and right brain regions is 84.20%and 83.07% under the condition of distinguishing brain regions and different music environments by extracting the wavelet energy entropy(WEE)feature of EEG signals as the input of the design optimized deep belief network(DBN)model.Compared with the classification accuracy of DBN,RBM and KNN in the mixed music environment,the classification effect is improved by 3.49%,12.89% and 7.24%.The data source of this paper is mainly EEG signals recorded in deap database,supplemented by self collected data,forming a comparative study in different music environments.Fast ICA,a fast algorithm of empirical mode decomposition(EMD)and independent component analysis(ICA),is combined to optimize the EEG signal,which can reduce the deep noise and remove the hybrid noise.The EEG signal characteristics of the left and right brain regions of the participants under different styles of music stimulation were extracted respectively,including: analysis of variance(ANOVA)in the time domain,power spectral density(PSD)of three brain wave rhythms related to music and emotion changes in the frequency domain,time-frequency-domain Hjorth parameters based on time series and wee in non-linear dynamics,so as to analyze the difference of EEG signal characteristics and study the sensitivity of left and right brain areas to different music.In the aspect of emotion classification and recognition,we use the optimized DBN model to realize the two classification of emotion in wake-up and potency dimensions,and study the laterality of left and right brain areas to emotion in different music environments.The conclusion of this study is of great significance for the diagnosis and treatment of patients with different brain injury and mental diseases in music therapy,and provides theoretical reference for stimulating different brain areas to stimulate directional emotions in medicine,so as to achieve better music assisted therapy effect.