내용보기

웨이브렛 계수와 퍼지 추론 시스템을 이용한 부정맥의 검출

초록/해제: 요약 :In this paper, we designed a fuzzy inference system to classify the premature ventricular contraction(PVC) based on the wavelet coefficients. Using power spectrum analysis, we observed that QRS of the PVC make a salient difference with that of the normal ECG at frequencies below 10Hz. In order to discern the PVC, it is very important to select the features which well describe the characteristics of the PVC. We chose three parameters - RR interval and two wavelet coefficients - as features. Firstly, we detected QRS using the dynamic threshold method and calculated the RR interval based on QRS position. The detected QRS was divided into several frequency bands by using Haar wavelet. Two features were extracted from the 6th wavelet coefficients(A6 and D6) among the several frequency bands. Finally, we designed a fuzzy inference system for the classification of the PVC using the three features. The MIT/BIH arrhythmia database including the PVC is used for the evaluation. The designed fuzzy inference system showed the PVC classification ratio of 96.09% and performed better than the neural network method. We showed the possibility that the fuzzy inference system can be used for the various ECG diagnosis.