Automatic detection of end QRS notching or slurring.

ABSTRACT

The purpose of this study was to define criteria suited to automated detection of end QRS notching and slurring and to evaluate their accuracy. One hundred resting 12 lead ECGs from young adult men, split randomly into equal training and test sets, were examined independently by two reviewers for the presence of such notching or slurring. Consensus was reached by re-examination. Logic was added to the Glasgow resting ECG program to automate the detection of the phenomenon. After training, the automated detection had a sensitivity (SE) of 92.1% and a specificity (SP) of 96.6%. For the test set, SE was 90.5%, SP 96.5%. Two populations of healthy subjects--one Caucasian, one Nigerian--were analysed using the automated method. The prevalence of notching/slurring with peak/onset amplitude respectively ≥ 0.1 mV in two contiguous inferolateral leads was 23% and 29% respectively. In conclusion, the detection of end QRS notching or slurring can be automated with a high degree of accuracy.