Application of the modernized wavelet transform to highlight the dynamics of changes in the duration of intervals during electrocardiogram diagnostics

Abstract

BACKGROUND: Cardiovascular diseases represent the leading cause of mortality worldwide [1]. A significant proportion of medical diagnoses are based on the evaluation of characteristic points in the electrocardiographic signal. For example, two important time intervals are P–R and Q–T, which have a significant impact on the patient’s health status [2]. However, the detection of minimal changes in amplitudes and intervals between waves over time is challenging through visual inspection alone. The difficulty is compounded by the lack of a clear-cut rule for determining the beginning and end of the Q–T interval, and the fact that the duration of the intervals varies with each heartbeat [3]. AIM: The study aimed to develop an algorithm to highlight the dynamics of interval duration changes when analyzing electrocardiographic signals. MATERIALS AND METHODS: The wavelet transform serves as a valuable analytical tool. Its ability to decompose signals into well-localized basis functions makes it well suited to distinguish electrocardiographic waves from noise [4]. Furthermore, its ability to change the scale allows for the detection of various local inhomogeneities in the electrocardiographic signal, as well as their durations. One of the main problems in using wavelet transform is the choice of the mother function. In this paper, we propose to use Hermite transform [5], due to which a mother function of arbitrary shape can be designed, which improves the detection efficiency. Moreover, the Hermite transform can be applied to the authentic electrocardiographic signal recording, ensuring the retention of the distinctive attributes of the patient’s signal. RESULTS: The result of the algorithm is a set of rhythmograms, each of which traces the changes over time of intervals of the electrocardiographic signal, for example, P–R or Q–T. The rhythmogram is a stochastic characteristic that allows estimation of the dispersion of Q–T intervals even during short time intervals and when changing the level of physical activity. This is why, by applying the statistical apparatus, it is possible to quantify the diagnostic efficiency of the proposed processing algorithm. CONCLUSIONS: The paper presents the main conclusions of the algorithm and the results of processing model electrocardiographic signals.