Abstract
In an automatic interpretation of electrocardiogram (ECG) by a digital computer, it is important to recognize the irregularity of rhythms and to read the waveforms of one particular cycle.
This paper deals with an automatic pattern recognition of the ECG rhythms based oz1 processing of 150 samples of clinical ECGs, each consisting of a series of 127 successive RR time intervals. There are many kinds of patterns of ECG rhythms, but in this paper, compensatory extrasystole, non-compensatory extrasystole, interpolated extrasystole, block, respiratory arrhythmia, bigeminy, trigeminy, polygeminy, Wenckebach's period, and several others were taken as fundamental types of rhythms. For the automatic recognition of these patterns of rhythms by a digital computer, firstly, quantitative definition of each pattern was given and then a special classification logic was constructed by means of deterministic, not by stochastic approach.
The results of computation indicate that the quantitative evalutation of rhythm as described in this paper provides a useful basis for the computer diagnosis of heart disease as well as of arrhythmia.
