Comparison of automatic QT measurement techniques in the normal 12 lead electrocardiogram.

OBJECTIVE: To undertake a quantitative assessment of different automatic QT measurement techniques and investigate the influence of electrocardiogram filtering and algorithm parameters. DESIGN: Four methods for identifying the end of the T wave were compared: (1) threshold crossing of the T wave (TH); (2) threshold crossing of the differential of the T wave (DTH); (3) intercept of an isoelectric level and the maximum T wave slope (SI); and (4) intercept of an isoelectric level and the line passing through the peak and the point of maximum slope of the T wave (PSI). Automatic QT measurements were made by all techniques following different electrocardiogram filtering and, when appropriate, with four different isoelectric levels and with three different threshold levels. SUBJECTS: 12 simultaneous standard electrocardiogram leads, containing at least two electrocardiogram complexes, were recorded from 25 healthy volunteers relaxing in a semirecumbent position. MAIN OUTCOME MEASURE: Mean and standard deviation of differences between reference and automatic QT measurements were compared for the four techniques. RESULTS: The mean automatic QT measurements varied by up to 62 ms, which was greater than has been found between manual measurements by experienced clinicians. Technique TH was particularly poor. The other techniques produced consistent results for most electrocardiogram filter, isoelectric level, and threshold level setting; but technique SI underestimated QT relative to the other techniques. CONCLUSION: Different QT measurement techniques produced results which were influenced, to varying degrees, by filtering and technique variables. This is relevant for the inter-comparison of studies using different techniques. Technique TH, a common approach, is not recommended.