Testing Gaussian-based kernels for modelling T-waves and P-waves in ECG signals

This paper presents a comparative study of segmentation and modelling of P and T waves in electrocardiograms, using three different mathematical models: Gaussian function, a composition of two Gaussian functions and Rayleigh probability density function. In order to evaluate the adaptability and the matching degree between each model and each characteristic wave, we compute the evolution of the corresponding parameters related to the fitted kernels throughout ECG records from the well-known QT database, as well as the normalized least mean square error between each model and the analysed waves. We have found the most accurate results for the kernel derived from the composition of two Gaussian functions, for which the average of normalized relative least mean square errors for T-wave and P-wave were, respectively, 1,56% and 9,13%, considering both available leads from the QT database.