A Study on Labeling Algorithm of ECG Signal using Fuzzy Clustering

퍼지 클러스터링을 이용한 심전도 신호의 구분 알고리즘에 관한 연구

  • 공인욱 (연세대학교 전기공학과) ;
  • 권혁제 (연세대학교 전기공학과) ;
  • 이정환 (연세대학교 전기공학과) ;
  • 이명호 (연세대학교 전기공학과)
  • Published : 1999.05.01

Abstract

This paper describes an ECG signal labeling algorithm based on fuzzy clustering, which is very useful to the automated ECG diagnosis. The existing labeling methods compares the crosscorrelations of each wave form using IF-THEN binary logic, which tends to recognize the same wave forms such as different things when the wave forms have a little morphological variation. To prevent this error, we have proposed as ECG signal labeling algorithm using fuzzy clustering. The center and the membership function of a cluster is calculated by a cluster validity function. The dominant cluster type is determined by RR interval, and the representative beat of each cluster is determined by MF (Membership Function). The problem of IF-THEN binary logic is solved by FCM (Fuzzy C-Means). The MF and the result of FCM can be effectively used in the automated fuzzy inference -ECG diagnosis.

Keywords