Noise stress testing for real-time ST segment measurement algorithms: a new methodology

A simple and computationally efficient testing methodology for quantitative performance evaluation of ST measurement algorithms in the presence of noise is described. Noise signals used include baseline wander, electrode motion artifact, and muscle artifact. Evaluation of testing results is done using graphical trends of the algorithm-generated ST measurements and histograms of the differences between the algorithm-generated ST values and the true ST values. Two new performance measures, the ST measurement sensitivity (STM-SE) and the ST measurement positive predictivity (STM-PP) are also defined to further quantify the performance results. Using this methodology, it is shown that the performance of a ST monitoring algorithm in the presence of noise can be evaluated and quantified