Correlation of nonlinear mathematical transformation of the normal electrocardiogram with the severity of coronary artery disease.

Signal processing was applied to normal ECGs to study whether the results could predict coronary artery disease (CAD) on angiography. ECGs determined to be normal by conventional criteria for 116 white males and 84 white females were subjected to signal processing. The technique applied nonlinear mathematical transformation to a segment of seven ECG leads which were photographically imaged, automatically digitized, normalized for time, and had topographical coordinate transformation using an IBM-XT microcomputer. The resulting curvilinear display on the computer monitor was termed a biopotential coordinate transformation (BCT). Successive superimposition of BCTs from normal ECGs of patients with normal angiograms developed boundaries of a normal template for each lead. Study BCTs were then compared to the normal templates to predict the presence or absence of CAD. Correlation of the BCT results with the severity of CAD for the males resulted in average sensitivities of 80%, 84.4%, and 91.7% for single, double, and triple-vessel CAD, respectively. Similarly, for the females average sensitivities were 59.1%, 73.9%, and 88.9%, respectively. In past studies on the same population, the BCT process detected CAD with average sensitivity of 85.1% and specificity of 81.8% for white males and average sensitivity of 73% and specificity 81% for white females. As such, computerized signal processing of ECGs may provide a noninvasive method for detection of CAD in patients with seemingly normal resting ECGs. The results of this study indicate that the sensitivity of the process increases in direct proportion to the severity of CAD. Further investigation is warranted for process refinement and verification.

Detection of coronary artery disease from the normal resting ECG using nonlinear mathematical transformation.

In order to enhance subtle changes not apparent in resting ECGs by conventional criteria, a computerized signal processing technique, biopotential coordinate transformation (BCT), has been developed to increase the physician's perceptibility of CAD in seemingly "normal" resting ECGs. ECGs initially read as normal then can be tested with this process to predict the presence or absence of CAD. A blinded, retrospective study of 93 Caucasian patients with normal ECGs was performed using coronary angiography as the test standard. The BCT process identified the presence or absence of CAD with an 84.3% sensitivity and 81.8% specificity in 62 men. Similarly, a 76.2% sensitivity and 80% specificity was obtained for 31 women. These results suggest that the BCT process may provide the emergency physician with a noninvasive screening test for the detection of CAD in patients with apparently normal resting ECGs. Further studies are necessary for process refinement and verification.

Nonlinear transformation of the resting electrocardiogram in the diagnosis of coronary artery disease.

Despite the common use of the standard 12-lead ECG, its reliability as an indicator of coronary artery disease (CAD) is poor. The normal ECG is falsely negative in more than 50% of angiographically proven CAD. The waveform of the standard ECG, however, can be transformed mathematically by nonlinear signal transformation to enhance its interpretation by computer. Using such a process it is possible that abnormalities can be identified in "normal" ECGs that can be correlated with CAD, thus identifying high-risk patients. A computer template that represents grouped data of normal ECGs for patients who also have normal coronary angiography was developed. Unblinded, preliminary testing of the template on normal ECGs of 107 white patients who had normal or abnormal coronary angiograms was performed. The process identified presence or absence of CAD with 82% specificity and 71% sensitivity for 53 women, and with 82% specificity and 86% sensitivity for 54 men. These preliminary results are promising, but further refinement of the templates is required and blinded studies with larger numbers and varieties of patients are needed.