Some theoretical results on the observability of repolarization heterogeneity on surface ECG

ECG-based data-driven solutions for diagnosis and prognosis of cardiovascular diseases: A systematic review

Cardiovascular diseases (CVD) are a leading cause of death globally, and result in significant morbidity and reduced quality of life. The electrocardiogram (ECG) plays a crucial role in CVD diagnosis, prognosis, and prevention; however, different challenges still remain, such as an increasing unmet demand for skilled cardiologists capable of accurately interpreting ECG. This leads to higher workload and potential diagnostic inaccuracies. Data-driven approaches, such as machine learning (ML) and deep learning (DL) have emerged to improve existing computer-assisted solutions and enhance physicians' ECG interpretation of the complex mechanisms underlying CVD. However, many ML and DL models used to detect ECG-based CVD suffer from a lack of explainability, bias, as well as ethical, legal, and societal implications (ELSI). Despite the critical importance of these Trustworthy Artificial Intelligence (AI) aspects, there is a lack of comprehensive literature reviews that examine the current trends in ECG-based solutions for CVD diagnosis or prognosis that use ML and DL models and address the Trustworthy AI requirements. This review aims to bridge this knowledge gap by providing a systematic review to undertake a holistic analysis across multiple dimensions of these data-driven models such as type of CVD addressed, dataset characteristics, data input modalities, ML and DL algorithms (with a focus on DL), and aspects of Trustworthy AI like explainability, bias and ethical considerations. Additionally, within the analyzed dimensions, various challenges are identified. To these, we provide concrete recommendations, equipping other researchers with valuable insights to understand the current state of the field comprehensively.

Dispersion of ventricular repolarization: Temporal and spatial

Repolarization heterogeneity (RH) is an intrinsic property of ventricular myocardium and the reason for T-wave formation on electrocardiogram (ECG). Exceeding the physiologically based RH level is associated with appearance of life-threatening ventricular arrhythmias and sudden cardiac death. In this regard, an accurate and comprehensive evaluation of the degree of RH parameters is of importance for assessment of heart state and arrhythmic risk. This review is devoted to comprehensive consideration of RH phenomena in terms of electrophysiological processes underlying RH, cardiac electric field formation during ventricular repolarization, as well as clinical significance of RH and its reflection on ECG parameters. The formation of transmural, apicobasal, left-to-right and anterior-posterior gradients of action potential durations and end of repolarization times resulting from the heterogenous distribution of repolarizing ion currents and action potential morphology throughout the heart ventricles, and the different sensitivity of myocardial cells in different ventricular regions to the action of pharmacological agents, temperature, frequency of stimulation, etc., are being discussed. The review is focused on the fact that RH has different aspects – temporal and spatial, global and local; ECG reflection of various RH aspects and their clinical significance are being discussed. Strategies for comprehensive assessment of ventricular RH using different ECG indices reflecting various RH aspects are presented.

A Potential-Based Inverse Spectral Method to Noninvasively Localize Discordant Distributions of Alternans on the Heart From the ECG

T-wave alternans (TWA), defined as the beat-to-beat alternation in amplitude of the T-waves, has been shown to be linked to ventricular fibrillation (VF). However, current TWA tests have high sensitivity but low specificity in determining who is at risk. To overcome this limitation, it might be helpful to determine the spatial distribution of any regions on the heart that alternate in opposite phase. Understanding these spatial distributions in relation to the regular activation of the heart could help explain the mechanism for the genesis of VF and thus disambiguate the low specificity of TWA.

GOAL

Image the spatial distribution of TWA on the heart surface from ECG measurements.

METHODS

We introduced the inverse spectral method (ISM), a tailored inverse (or ElectroCardioGraphic Imaging) solution designed specifically to noninvasively image cases of TWA on the heart.

RESULTS

We evaluate the ISM on its capacity to reliably detect the spatial distributions of TWA compared against a standard TWA detection method applied directly to the electrograms on the heart surface. We report on results from both a series of synthetic simulations of TWA generated using the ECGSIM software and a set of continuous epicardial surface voltage recordings from a canine experiment. ISM detected TWA distributions that matched the phase of the true underlying out-of-phase regions over and of the heart surface, respectively.

CONCLUSION

Our results suggest that ISM is capable of reliably detecting the different regions present in a TWA distribution across a wide variety of TWA locations on the heart in simulation and in the face of transients and nonidealities in the canine recordings.

Effect of action potential duration on Tpeak-Tend interval, T-wave area and T-wave amplitude as indices of dispersion of repolarization: Theoretical and simulation study in the rabbit heart

BACKGROUND

The aim of the study was to differentiate the effect of dispersion of repolarization (DOR) and action potential duration (APD) on T-wave parameters being considered as indices of DOR, namely, Tpeak-Tend interval, T-wave amplitude and T-wave area.

METHODS

T-wave was simulated in a wide physiological range of DOR and APD using a realistic rabbit model based on experimental data. A simplified mathematical formulation of T-wave formation was conducted.

RESULTS

Both the simulations and the mathematical formulation showed that Tpeak-Tend interval and T-wave area are linearly proportional to DOR irrespectively of APD range, while T-wave amplitude is non-linearly proportional to DOR and inversely proportional to the minimal repolarization time, or minimal APD value.

CONCLUSION

Tpeak-Tend interval and T-wave area are the most accurate DOR indices independent of APD. T-wave amplitude can be considered as an index of DOR when the level of APD is taken into account.

Spatial repolarization heterogeneity and survival in Chagas disease

OBJECTIVES

We investigated if cardiac spatial repolarization heterogeneity might be associated with an increased risk of death in patients with chronic Chagas disease.

METHODS

Repolarization heterogeneity was assessed using the V-index, a recently introduced metric founded on a biophysical model of the ECG. This metric provides an estimate of the standard deviation of the repolarization times across the heart. We analyzed 113 patients (aged 21- 67 years) enrolled between 1998 and 1999 who had a known serological status showing positive reactions to Trypanosoma cruzi. Fourteen subjects died during a 10-year follow-up period.

RESULTS

The V-index was significantly lower in survivor (S) than in non-survivor (NS) subjects (S: 31.2 ± 13.3 ms vs NS: 41.2 ± 18.6 ms, single-tail t-test: p = 0.009, single-tail Wilcoxon rank sum test: p = 0.029). A V-index larger than 36.3 ms was related to a significantly higher risk of death in a univariate Cox proportional-hazards analysis (hazard ratio, HR = 5.34, p = 0.0046). In addition, V-index > 36.3 ms retained its prognostic value in a multivariate Cox proportional-hazards analysis after adjustment for other three clinical variables (left ventricular ejection factor < 0.50, QRS duration > 133 ms, ventricular tachycardia during stress testing or 24 hours Holter) and for T-wave amplitude variability > 30 μV, even using shrinkage, a statistical procedure that protects against over-fitting due to small sample size.

CONCLUSIONS

The study showed that an increased dispersion of repolarization times in patients with Chagas disease, as measured by the V-index, is significantly correlated with the risk of death in a univariate survival analysis. The V-index captures prognostic information not immediately available from the analysis of other established risk factors.