Research On Accurate Pattern Recognition Of Bruxism Based On Surface Electromyography

Bruxism is a kind of oral disease,which has an adverse impact on patients' health.The patients' teeth are seriously damaged during the long process of teeth clenching and grinding.At present,preliminary judgment can only be made according to the degree of tooth wear whereas accurate identification of bruxism remains a challenging task for dentists.Therefore,it is of significance to develop an accurate and safe technology for identifying bruxing patterns,which can assess disease status of bruxism quantitatively and effectively.In this paper,surface electromyography(sEMG)of masticatory muscle was collected simultaneously during teeth clenching or grinding and further analyzed.sEMG were investigated for bruxing pattern identification,and various characteristic parameters,including the intensity and frequency of bruxism,were obtained,which could pave the way for future clinical application.The main work and innovations of this paper are as follows:1.Since bruxism is accompanied with nonfunctional contraction of masticatory muscles,a sEMG detection system was designed to detect the sEMG signals during bruxism.The system hardware and corresponding embedded software design were accomplished,allowing for basic functions including sEMG collection,storage and display.Pattern recognition function and characteristic parameter calculation of bruxism were also integrated in this system.2.The relationship between intensity of teeth clenching and sEMG was studied,and accurate measurement of teeth clenching intensity was achieved by sEMG.Intensity of teeth clenching measured by thin film pressure sensor and sEMG were collected simultaneously,and their mathematical relationship was acquired.It was shown that,sEMG of masticatory muscle and intensity of teeth clenching demonstrated identical changing tendency,and thus sEMG could be an effective indicator reflecting the intensity of teeth clenching.3.Characteristic parameters of sEMG were extracted in time domain,frequency domain and entropy value,including Root Mean Square(RMS),Integral Electromyography(iEMG),Median Frequency(MF),Mean Power Frequency(MPF),Approximate Entropy(ApEn),and Sample Entropy(SampEn).Based on analysis results,ApEn and SampEn were eventrually determined as the most effective characteristic parameters for accurate pattern recognition of bruxism.4.Pattern recognition for bruxism was obtained based on the analysis of sEMG.Three bruxing patterns and corresponding parameters can be identified including resting state,clenching state and grinding state.Charateristic parameters of bruxism were extracted from sEMG towards quantitative assessment of bruxism.Then cluster centers for training sets were calculated and feature vectors of testing sample were extracted.Mahalanobis distances between cluster centers and feature vectors were calculated and compared to achieve accurate pattern recognition.It was shown that accurate rate of clenching state recognition was more than 90%,and accurate rate of resting state and grinding state recognition were more than 85%.The intensity,frequency and time duration of bruxism can be accurately determined accordingly.In this paper,the accurate pattern recognition and quantitative assessment of bruxism was realized.This study owns clinical relevant potentials for future diagnosis and treatment evaluation of bruxism.