The effects of noise on computerized electrocardiogram measurements

Computerized measurements provide objective and reproducible assessments of the electrocardiogram (ECG). These measurements may be affected by noise or other lead quality issues. The effects of noise on the repeatability of computer-measured PR interval, QRS duration, QT interval, P/QRS/T axes, and ST levels were examined.MethodsThe 125 ECGs of the Common Standards for Quantitative Electrocardiography (CSE) measurement database (MO1 series) were merged with records from the MIT Noise Stress Test database. For each CSE ECG, 720 unique noise ECGs were created, for a total of 90 000 noisy ECGs. Computerized measurements from the noisy ECGs were compared to the original ECG measurements. The repeatability of the measurements was assessed as a function of a lead quality score.ResultsThe repeatability of the measurements was found to be in excellent agreement with the original ECG measurements when the noise level was no worse than that of the original ECGs. Noise did not introduce any bias to the measurements, although not surprisingly, the variation of the errors increased as the lead quality degraded.