M. L. Riccio, N. S. Moise, N. F. Otani, J. C. Belina, A. R. M. Gelzer, and R. F. Gilmour, Jr., Vector Quantization of T Wave Abnormalities with a Predisposition to Ventricular Arrhythmias and Sudden Death, Annals of Noninvasive Electrocardiology 3, 46-53 (1998).

Abstract

Background: Previous studies have suggested that irregular T wave morphologies are associated with an increased risk of sudden death. However, automated characterization of T wave abnormalities has been hampered by the lack of suitable analysis techniques. In this study, we tested a new method of T wave analysis in German shepherd dogs with inherited ventricular arrhythmias and sudden death.
Methods: Sets of 24-hour ambulatory ECG recordings obtained from unaffected (n=6) and affected (n=5) dogs were digitized, automatically annotated to label each R wave, and placed into a matrix with the R waves aligned. A vector quantization algorithm separated the QRS-T complexes into classes according to T wave morphology. The existence of notched T waves was determined by assessing the number of zero crossings of the first derivative during the T wave.
Results: The duration of the QT interval was similar in affected and unaffected dogs (182 +- 14 ms vs 176 +- 16 ms, respectively). However, T wave morphology differed between the two groups. Specifically, affected dogs had a higher percentage of notched T waves than unaffected dogs (41.6% +- 10.8% vs 5.0% +- 1.2%, respectively). Notched T waves did not appear at all times of day, nor were they present in all leads.
Conclusion: Vector quantization and first derivative analyses were feasible and effective methods for detecting T wave abnormalities associated with the development of ventricular arrhythmias. These methods ultimately may be useful for risk stratification of patients susceptible to ventricular arrhythmias and sudden death.