A Fuzzy Logic-Controlled Classifier for Use in Implantable Cardioverter Defibrillators JODIE USHER. DUNCAN CAMPBELL, JITU VOHRA,* and JIM CAMERON From the Department of Electronic Engineering, Biomedical Engineering, La Trobe University>&nd *Royal Melbourne Hospital, Melbourne, Victoria, Australia —' USHER J.. ET AL.: A Fuzzy Logic-Controlled Classifier for Use in Implantable Cardioverter Defibrillators. Purpose: Implantable cardioverters defibrillators (ICDs) are increasingly ustid in the management of life- threatening arrhythmias. Correct recognition of a treatable arrhythmia is crucial to this application. How- ever, the computational power of microprocessors currently used in ICDs limits the range of traditional al- gorithms available for this application.Methods: Classification based on fuzzy inference systems (FIS) were trained to recognize different cardiac rhythms (AF, VF, SVT, VT) from the Ann Arbor Electrogram Library. The FIS used were designed using adaptive-network-based fuzzy inference methods to optimize the classification procedure. Only computational techniques suitable for ICD design were used. Results: After pretraining with the ANFIS correct rhythm classification was observed for the rhythms studied. Con- clusion: In this preliminary study, successful rhythm classification was demonstrated using fuzzy logic techniques. In view of the computational efficiency this may have application in ICD design. (PACE 1999; 22[Pt. 111:183-186} implantable cardioverter defibrillator, fuzzy logic, rhythm classification Introduction Differentiation of cardiac rhythms using au- tocorrelation techniques is widely used in the analysis of surface ECGs. However, due to its high computational demand, it is not suitable for im- plementation in the microprocessors currently availahle in implantable cardioverter defibrilla- tors (ICDs). Various techniques have been pro- posed to estimate the normalized autocorrelation of electrograms with less computational require- ments. However, these have in general remained too complex for ICD use. The objective of the work presented here was to investigate the suit- ability of fuzzy logic techniques (as initially de- scribed by Zadeh in 1963*) to clarify cardiac ar- rhythmias with a view to their potential application in ICD technology. A nonlinear pre- dictor using adaptive network-based fuzzy inter- face systems (ANFIS) was used to clarify the ar- rhythmias tested.^ Address for reprints: James Cameron, M.D., Department of Electronic Engineering, La Trobe University, Plenty Road, Bundoora, Victoria, Australia 3083. Fax: +61-0-3-9471-0524; e-mail: j.cameronfaee.tatrobe.edu.au Methods The results presented here were obtained us- ing input data sets from the Ann Arbor Electrogram Library (Ann Arbor, MI, USA) and Matlab analysis software version 4.2cl (The MathWorks Inc.. Nat- ick, MA, USA) with the associated Fuzzy Logic Toolbox.^ Electrogram sampling rate was 100 Hz and the first 2,000 points of each recording was used to train the system with the remainder of the record used to test rhythm recognition. The right ventricular apical electrogram was used in all tests. One Sugeno-type fuzzy inference system (FIS)* is trained to predict subsequent data points for each rhythm likely to he encountered (in this study SVT, VT, AF, VF) using an ANFIS proce- dure.^ Each FIS was arbitrarily specified to consist of two bell-shaped membership functions, each of the general form •4 f[x;a,/3,y)= — 1 + X — a where a defines the center of the curve, p the half- amplitude half-width, and -y the sharpness of the PACE, Vol. 22 January 1999, Part 183