Computer analysis of exercise-induced changes in electrocardiographic variables. Comparison of methods and criteria

In order to evaluate computerized methods of electrocardiographic signal processing, determination of QRS end, and measurement of criteria for ischemia, we analyzed the data from 42 male patients with coronary heart disease who underwent maximal treadmill testing. Electrocardiographic data were digitized on-line and leads X, V5, Y and Eigen V were later analyzed for noise content, isoelectric baseline, and ST parameters using the UCSD spatial electrocardiographic computer program. Various ST segment criteria for ischemia were calculated and compared. Noise was greater in lead Y and in all leads when the median was used for signal averaging. Two isoelectric baseline algorithms and three ST segment slope algorithms gave similar results. Spatially derived QRS end was highly correlated with the amplitude measured using a fixed time interval after peak R wave. Both ST area and ST midpoint estimates differed widely using two different algorithms for each. Regression equations were derived that make it possible to estimate QRS end or ST60 amplitudes in V5 from values in X or vice versa.