A signal-to-noise ratio based false identification reduction system was proposed for an ECG based biometric system. The system generated a signal quality index (SQI) based on the 25th percentile of the signal-to-noise ratios of the individual beats within a 20 s segment of ECG data. Identifications generated from ECG segments with SQIs below a set threshold were gated. The system was tested using 642 ECG segments collected from 32 subjects while standing still and while jogging. With no gating the biometric system attained a precision of 0.49. Following the application of the gating system at a threshold of 1 dB, the precision increased to 0.68. The system eliminated 98.7% (155/157) of the false identification during the noise corrupted (jogging) interval while maintaining the count of the true identifications (2/2). During the clean (standing still) intervals, the system gated 57.8% (193/334) of the false identifications and 8.14% (25/307) of the true identifications.

Additional Metadata
Keywords biometric, continuous identification, electrocardiogram, false identification, motion artifact, Signal Quality Index, signal to noise ratio
Persistent URL dx.doi.org/10.1109/MeMeA.2017.7985899
Conference 12th IEEE International Symposium on Medical Measurements and Applications, MeMeA 2017
Citation
Abdelazez, M. (Mohamed), Hozayn, M. (Mohamed), Hanna, G.S.K. (George S.K), & Chan, A. (2017). Gating of false identifications in electrocardiogram based biometric system. In 2017 IEEE International Symposium on Medical Measurements and Applications, MeMeA 2017 - Proceedings (pp. 338–343). doi:10.1109/MeMeA.2017.7985899