Exercise stress tests for evaluation of ischemic heart disease and functional performance of the heart have been used with increasing frequency in recent years. This increase is due to several factors. One of which is the need to objectively evaluate evidence of myocardial ischemia for correction by coronary artery saphenous vein bypass grafting. An additional factor is the evidence that stress testing is a reproducible method for functional evaluation of the patient's performance. Many criteria have been proposed for evaluating the electrocardiogram (EKG) taken during and following exercise to determine the presence of ischemic heart disease (2, 3, 5, 6). Common to all of these criteria, however, is the inability to record adequate signals during exercise. In addition, the volume of data collected requires considerable physician time if adequate interpretation is to be achieved.
This paper describes an on-line computer program developed at the Latter-day Saints Hospital for processing of EKG signals before, during and after treadmill stress testing. The development of these programs began with a number of prerequisites. These requirements included : (1) that the data be collected and processed in an on-line, real-time process; (2) that use of the computer did not endanger the patient's safety; (3) that adequate back-up be present in case of computer failure; (4) that the testing procedure is not prolonged; (5) that the number of personnel and abilities of the assistants performing the test would not be increased by the addition of computer analysis; and (6) that computer processed data provide distortion free signals,
Signal noise during exercise test recordings fall into two categories, the first being baseline wander due to impedance changes in the electrode-skin interface, and the second being muscle artifact. Figure 1 is a recording of an EKG taken on a patient at this hospital during a maximum exercise test. Because of the magnitude of the muscle artifact superimposed on this electrocardiogram, interpretation of the tracing is very difficult. In some points it is almost impossible. Since the noise is present regardless of the particular criteria for determination of normal or abnormal * Supported in part by NIH Grant No. RR-00012.