Software techniques for solving two problems in monitoring the electrocardiograms of critically ill patients are described in this Paper.
(1) Logical filter for detection of a QRS complex : High frequency components are centered around the peak-points of a QRS complex and the waveform is roughly symmetrical with respect to a vertical line through.the peak-point. These characteristics of the QRS complex are utilized in designing the filter. This filter distinguishes the QRS complex well from other parts of electrocardiograms and various noises.
(2) Pattern recognition of a QRS complex with correlation coefficient : One typical QRS complex is selected and designated as “templet”. Subsequent QRS complexes are compared with this “templet” one by one. The degree of similarity to the “templet” is evaluated by a correlation coefficient. Without accurate detection of a QRS duration, 10 points before and after the peak of R-wave respectively (totalling 20 points) are regarded as one QRS complex (sampling rate : 250 points/second). Our method makes it possible to recognize ectopic beats automatically even if noises exist to some extent.