Abstract
from the observation that techniques traditionally used to enhance the signal-to-noise ratio of the QRS, can, with a simple window change, also generate adaptive thresholds that are low enough to detect almost every QRS, regardless of amplitude or morphology, yet high enough to disregard artefact and irrelevant fluctuations in the signal. MAST is additionally able to automatically annotate and exclude segments of the signal where QRS detection is impossible due to artefact levels exceeding signal levels. RESULTS: Two expert reviewers compared the performance of MAST, in terms of true positive, false positive and false negative detections, with that of popular commercial ECG analysis software, Dataquest ART (Version 4.1, Data Sciences International, St. Paul, MN, 2007). Both algorithms were applied blindly to the same random selection of 270 three-minute ECG recordings containing a total of 437,533 manually-annotated QRS complexes and varying amounts of signal artefact. The error rate exhibited by MAST was roughly a quarter that of Dataquest ART, with significantly greater consistency and virtually no false positives (Se: 98.48% 4.32% vs. 94.59% 17.52%, Sp: 99.99% 0.06% vs. 99.57% 3.91%, P 0.001 and P 0.027 respectively, Wilcoxon signed-rank test). CONCLUSION: MAST would be an asset to any investigator performing cardiovascular research in the murine model. It is computationally efficient and could be applied to other noisy environments such as long term ECG monitoring and exercise testing. Our preliminary attempts to adapt MAST to human electrocardiograms have resulted in a 99.7% detection accuracy on a database of Holter recordings with moderate to high noise.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have