The Research Of Acoustic Emission Signals Diagnose Knee Osteoarthritis

Aiming at the limitation of the high incidence of knee Osteoarthritis (short for OA), the traditionaldiagnosis methods lack of portability, expensive price, radiation damage, ect.The paper presents amethod to diagnose knee OA that make use of Acoustic Emission(short for AE) signal. The paper of thestudy includes the study of AE signal acquisition system,study of AE signal denoising method,researchof Intelligent diagnosis method,the correlation between knee osteoarthritis and pathogenic factors.Thepaper combined with sensor technology, computer technology, communication technology, thedenoising technology and intelligent diagnosis technology.Knee joint AE signal acquisition system consists of upper order computer system and lower ordercomputer system.The hypogyny machine system includes AE sensors and signal conditioningmodule.The combination of the two modules realized the acquisition of knee AE signal,then transmitthe collected data to PC by USB.PC system mainly implements the parameter settings, waveformdisplay, data display,etc.Convenient for preprocessing of subsequent data.After studying the knee joint’s characteristics of the AE signal，this paper first introduces the FastFourierTransform(FFT),and using FFT to denoising,then analyzes the advantages and disadvantages ofall kinds of wavelet base,select Sym8as the optimum wavelet base.To calculate the maximumdecomposition scale,using wavelet analysis to denoise.Compared with FFT de-noising.The denoisingeffect shows that using wavelet de-noising effect is better than that of FFT.Finally, the characteristicparameters were normalized.The paper studies the relationship between amplitude peak, average signal level (ASL),durationof knee joint of AE signal and six kinds of pathogenic factors of age, gender, heredity, obesity, traumaand force, bone mineral density in the the knee osteoarthritis.Compared the difference of the AEsignal between different age, gender, heredity, obesity, trauma and force, the differences of bone mineraldensity. Analysis results show that the six pathogenic factors are certain correlation with kneeosteoarthritis. Work curve using ROC curve evaluated amplitude peak, ASL, duration have good effectsin the diagnosis of knee OA.The data for intelligent diagnosis, making healthy group, slight knee osteoarthritis, severe kneeosteoarthritis as the research object.Choosing the BP neural network, genetic algorithm to optimize theBP neural network and support vector machine (SVM) for the diagnosis.Results show that using threediagnosis algorithm can diagnosis results, and the diagnosis effect is accurate, by comparison,classification of SVM is more quick, and high accuracy.The results of experiment of knee OA’s AE database established by the laboratory shows that the diagnostic method has well effect, small error, clinical value and to be able to real-time detection andearly prediction of knee OA patients.