Title :
Real-time ventricular arrhythmia detection with Fourier analysis and neural network
Author :
Minami, K. ; Ohkuma, Y. ; Nakajima, H. ; Toyoshima, T.
Author_Institution :
Medtronic Japan, Tokyo, Japan
Abstract :
We developed a system which detects life threatening ventricular arrhythmias with respect to each beat. In this paper, we applied the system to human intracardiac electrogram (EGM) data for demonstrating the clinical potential. The system analyzes Fourier spectrum of the EGM signal corresponding to an individual QRS complex, and classifies it to three kinds of rhythm origins using a neural network. In our study, supra-ventricular rhythms were classified from ventricle originated rhythms with high sensitivities and specificities.
Keywords :
Fourier analysis; backpropagation; electrocardiography; feature extraction; medical signal processing; patient monitoring; pattern classification; real-time systems; spectral analysis; EGM; Fourier analysis; beat; clinical potential; high sensitivities; high specificities; human intracardiac electrogram; individual QRS complex; life threatening ventricular arrhythmias; neural network; real-time ventricular arrhythmia detection; rhythm origins; supra-ventricular rhythms; ventricle originated rhythms; Animals; Electric shock; Failure analysis; Fibrillation; Humans; Neural networks; Rhythm; Shape; Timing; Virtual reality;
Conference_Titel :
Computers in Cardiology, 1996
Conference_Location :
Indianapolis, IN, USA
Print_ISBN :
0-7803-3710-7
DOI :
10.1109/CIC.1996.542594
