Validation of an algorithm to reveal the U wave in atrial fibrillation

Abstract

Major cardiac organisations recommend U wave abnormalities should be reported during ECG interpretation. However, U waves cannot be measured in patients with atrial fibrillation (AF) due to the obscuring fibrillatory wave. The aim was to validate a U wave measurement algorithm for AF patients. Multi-beat averaging was applied to ECGs of 25 patients during paroxysms of AF and the presence of U waves compared to those from the same patients during sinus rhythm (SR). In a further database of 10 long-term AF recordings, the number of beats for effective U wave extraction by the algorithm was calculated. U waves were revealed in all AF recordings and there was no significant difference between the presence of U waves in AF and SR (p = 0.88). U wave amplitude was significantly increased in AF (mean (s.d.) amplitude 55 (39) AF vs 37 (28) μV SR, p = 0.005). The presence of U waves could easily be discerned when as few as 10 beats were used in the algorithm. The study demonstrates the validity of the algorithm to reveal U waves in AF recordings. The algorithm offers the potential to detect U wave abnormalities in patients with AF.

Multi-Beat Averaging Reveals U Waves Are Ubiquitous and Standing Tall at Elevated Heart Rates Following Exercise

“…Exercise ECG is usually contaminated by noise, making it difficult to observe small amplitude features such as U waves. In our effort to facilitate the measurement of U waves in challenging conditions we have previously demonstrated ventricular beat averaging as an effective tool to reveal U waves when they are completely obscured by the fibrillatory wave in atrial fibrillation [ 5 ]. We hypothesise that the same algorithm can be applied to recover U waves from ECGs contaminated by noise due to exercise.…”

“…The algorithm uses signal averaging, which is a well-known signal processing technique used extensively to reduce noise when multiple repetitions of a small amplitude signal are available for measurement [ 5 , 11 , 12 ]. It allows small amplitude signal components to be measured in the presence of large amplitude noise, but only when the signal and noise are uncorrelated [ 11 , 12 , 13 ].…”

“…By careful alignment and averaging of increasing numbers of ventricular beats, the amplitude of the uncorrelated noise is progressively reduced while the desired U wave is revealed in the resulting average beat. The level of noise reduction achieved is a function of the number of averaged signal epochs ( N ) where it can be reduced by a factor of √ N [ 5 , 11 , 12 , 13 ].…”

“…U waves are difficult to measure due to their low amplitude, and contaminating noise or other ECG features may often obscure them. For example, U waves are obscured by the atrial fibrillatory wave during atrial fibrillation [ 5 ]. Similarly, at high heart rates the proceeding P wave may obscure the U wave [ 6 ].…”