A technique for avoiding false acceptance in ECG identification

The purpose of this study was to evaluate a new identification engine for an ECG identification system. This system identifies subjects based on a comparison of a measured ECG with previously registered ECG feature parameters. These feature parameters are sampled from the intervals and amplitudes of the electrocardiographic waves extracted using characteristic points appearing on the waveform of the second order derivative. Although discriminant analysis that has been used for the comparison shows sufficient performance, the system identifies any unregistered subject as registered. In order to avoid this problem, which is called false acceptance, we propose a technique that can classify ECG as unregistered or registered, using Maharanobis' generalized distance and threshold. The results of an experiment with 20 registered ECGs and 1 unregistered ECG show that false acceptance was avoided, and that proper threshold should be determined for lower false acceptance rate and acceptable false rejection rate.