Analysis And Identification Of EEG Signals Under Pure Tone Stimulation

Auditory is an important bridge for communication between people and the external world.It is clinically meaningful to detect and assess the degree of hearing impairment for hearing-impaired people.Basically,hearing is a behavioral responsing to sound,and frequency and intensity are important characteristics of hearing process.Investigating the processing of pure tones at different frequencies has clinical value in the evaluation of hearing impairment and auditory cognitive abilities.Pure tone audiometry,as an important valuation of hearing abilities will be affected age,education level and camouflage,leaving errors in the final test.It is referential to analyze the feature for pure tone audiometry.According to previous researches on the recognition of EEG signals induced by audio-visual and psychology.In this paper,firstly,EEG signals were collected from 24 subjects presented by six pure frequency stimulations.Secondly,the feature extraction and selection were carried out based on EEG.Thirdly,EEG signals were classified under different frequency pure tones,and the correlation of EEG signals and its frequency was explored.The specific work is as follows:(1)Designing an affective induction experiment to acquire the EEG signals induced by pure tones of six frequencies.The MATLAB generated 70 dB SPL six frequencies of pure tone(200Hz,400 Hz,800Hz,1600 Hz,3200Hz,6400Hz),using 32-lead caps to collect the EEG data generated by the pure tone stimulation at six frequencies.(2)Preprocessing the primitive data in order to obtain non-polluting EEG signals induced by six frequencies of pure tone.The analysis was referred to the preconditioning process of EEG signals in cognitive neuroscience paradigm.The artifacts of functional magnetic resonance were removed by EEGLAB and Analyzer software.Re-reference,downsampling,filtering,Removing were done to the artifact,such as EMG,EOG,and cooperate with the manual,to be perfect to obtain ideal EEG data.(3)Analyzing EEG signals induced by six frequencies of pure tone with timefrequency analysis.With reference to the EEGLAB tutorial,event-related spectral perturbation(ERSP)analysis is performed on the EEGLAB signals at six frequencies.We observed the change of the rhythm of EEG induced by pure tone over time,explore the mechanism of brain activity stimulated by pure tone,and compare the results under different conditions.The results showed that compared with the quiet state,delta(1-4Hz),theta(5-7Hz),alpha(8-12Hz)frequency stimulated by pure tones increased and showed difference among tones of different frequencies.(4)Feature extraction and selection of EEG signals induced by six frequencies of pure tone.This analyse referenced signal analysis theory,composition of auditory evoked potential and foundation of general signal.These features were extracted basing on nonstationary,strong randomness,non-linear and frequency domain features obvious features of EEGs,such as mathematical statistics(peak,variance),N1,P1,power spectrum estimate,energy,and approximate entropy characteristics.The back-selection algorithm was used for feature selection combining with the statistical T-test.The approximate entropy and frequency band energy were chosen for the better combination of features.(5)Classification and identification of six frequencies of pure tone.Using the two typical classifiers of support vector machine(SVM)and BP artificial neural network(ANN),and the average recognition rate based on approximate entropy is above 55%,and the best is 70.83%.The average recognition rate of SVM is batter than the BPANN.In this paper,we found that the brain responds to different frequency pure stimuli,and that the delta and theta frequencies of the brain are significantly increased with pure tone,and there are certain differences in the six frequencies in this frequency band.In general,EEG signals are distinguishable from the six frequencies of pure tone.As the frequency difference increases,the average recognition rate will increase.The research results of this paper have certain potential value for pure tone audiometry,humancomputer interface and blind person guidance.