Real-time implementation of twelve-lead automated electrocardiogram system measurement for QT dispersion analysis

Abstract

Research and study of the electrocardiogram evolves with the advancement of digital signal processing and artificial intelligence. Unfortunately, readily available electrocardiogram machines in the market do not provide automated measurement of the QT dispersion. Therefore, a twelve-lead electrocardiogram system is developed in order to assist the cardiologists in carrying out their research on the cardiac diseases. The development of the system consists of several phases. The first phase includes the construction of a real-time twelve channels data acquisition unit with the universal serial bus (USB) interface. The following phase includes the study and development of the electrocardiogram signal conditioning circuits. The third phase is the study of the designed anti-aliasing filters and its effect to the electrocardiogram distortion. The Butterworth and the Bessel filters, each with orders of two four and eight are compared and the 8th order Bessel low pass filter appears to be the best candidate. The subsequent phase is the implementation of the time-domain subtraction technique to remove the power line noise in the electrocardiogram signal with minimal distortion. The filter is compared to a notch Twin-T filter, and results showed that not only the time-domain subtraction technique suppresses noise, it also preserves the original signal with minimal distortion. The automated QT interval measurement algorithm is validated upon an annotated standard database, the Physikalisch-Technische Bundesanstalt (PTB) Diagnostic Electrocardiogram database which is being the focused for the International QT Interval Challenge 2006. Result shows that 28.53% of the database is correctly identified for the QRS onset and T offset locations due to the dissimilar morphologies of the electrocardiogram signal.