The study demonstrates the evaluation of two adaptive QRS detectors using ROC analysis on a database of difficult ECGs.
QRS detectors are often evaluated in terms of statistical measures, e.g., rate of false and true detections. These measures are usually calculated for a detector with fixed parameter values. In this paper, detector properties are studied by varying those parameters which seem to have the most profound effect on performance. By applying techniques known from optimal estimation to a stochastic model for the ECG, two different detectors have been derived. The detector structures are described in detail. In order to study and compare the performance of the two detectors, a database of "difficult" ECG's has been collected, containing various "artifacts" of physiological and electrical/mechanical origin. The performance of this database is presented in terms of receiver-operating characteristics (ROC's), and several examples are discussed.
Sörnmo et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: