Classification of the long-QT syndrome based on discriminant analysis of T-wave morphology

Abstract

The long QT syndrome (LQTS) is a genetic disorder, typically characterized by a prolonged QT interval in the ECG due to abnormal cardiac repolarization. LQTS may lead to syncopal episodes and sudden cardiac death. Various parameters based on T-wave morphology, as well as the QT interval itself have been shown to be useful discriminators, but no single ECG parameter has been sufficient to solve the diagnostic problem. In this study we present a method for discrimination among persons with a normal genotype and those with mutations in the KCNQ1 (KvLQT1 or LQT1) and KCNH2 (HERG or LQT2) genes on the basis of parameters describing T-wave morphology in terms of duration, asymmetry, flatness and amplitude. Discriminant analyses based on 4 or 5 parameters both resulted in perfect discrimination in a learning set of 36 subjects. In both cases cross-validation of the resulting classifiers showed no misclassifications either.

Appendix

The mean, width, asymmetry and flatness of the T-wave are calculated according to

$$ m_{p} = {\left[ {{\sum\limits_{n = 0}^{N - 1} {{\left( {t_{n} - {\text{ref}}} \right)}^{p} \cdot {\text{ecg}}(t_{n} )} }} \right]}^{{1/p}} $$

(2)

for p = 1, 2, 3, and 4, respectively, and where ref is the reference (ref = 0 for calculation of m ₁ ). The area of the T-wave was normalized before the m _p was calculated (Table 4).

Table 4 List of parameters to characterize the T-wave