Prediction of Ventricular Fibrillation in Myocardial Infarction Model in Pigs Based on T-peak-to-T-end Interval Variation Using Time Warping Technique

Background Introduction

Source of the Paper

Sudden Cardiac Death (SCD) is a leading cause of mortality worldwide, with one of its primary pathogenic mechanisms being Ventricular Fibrillation (VF), especially in the context of myocardial infarction. In this context, early prediction of VF risk becomes particularly important. The connection between changes in Ventricular Repolarization (VR) and the formation of ventricular arrhythmias has been validated in experimental models and clinical studies. The T-peak-to-T-end interval (Tpe) has been proposed as a substitute indicator for VR Dispersion (VRD) and is considered a potential predictor of arrhythmia risk. However, the Tpe interval only captures the time difference between the peak and end of the T wave and does not take into account the information contained within the waveform. Thus, this study quantifies the morphological changes within the Tpe interval through a newly proposed Time Warping-based metric to predict the occurrence of VF.

This study was co-authored by Neurys Gómez, Julia Ramírez, Alba Martín-Yebra, Marina M. Demidova, Pyotr Platonov, Juan Pablo Martínez, and Pablo Laguna (IEEE Members). The research institutions include the University of Zaragoza in Spain, Lund University in Sweden, and Queen Mary University of London in the UK. The paper was published in IEEE Transactions on Biomedical Engineering in 2024.

Research Objective

The study aims to analyze the temporal trajectory of the Tpe morphological change index based on time warping (dpca_w,tpe) during ischemia and to test its ability to predict imminent VF in a pig myocardial infarction model.

Research Methods

Experimental Dataset and Protocol

This study utilized a closed-chest pig myocardial infarction model implemented at Lund University in Sweden, introducing a 40-minute coronary artery occlusion experiment. The subjects were 32 domestic pigs, and detailed analyses were conducted on their ECG records before and during the coronary occlusion.

ECG Preprocessing

First, the ECG signals were filtered using a bidirectional sixth-order Butterworth low-pass filter and a high-pass filter to remove electrical noise and baseline drift, respectively. Next, a wavelet-based single-lead ECG delineation method was applied to determine the characteristic points of the QRS complex and transformed through Principal Component Analysis (PCA) to emphasize the T wave content.

Time Warping to Quantify Tpe Interval Changes

For each pig, the dynamic change in the Tpe interval was quantified through Time Warping. The T-peak-to-T-end (Tpe) interval was re-parameterized in time using the method proposed by Ramírez et al., calculating the warping index dpca_w,tpe.

Statistical Analysis

Mann-Whitney tests were used to compare the index values between the non-VF and delayed VF groups. The predictive capability over different time intervals was analyzed using Receiver Operating Characteristic (ROC) curves.

Research Results

Changes in DPCA_W,TPE During Occlusion

The experiment showed that the dpca_w,tpe index remained stable at baseline but gradually increased during coronary occlusion, particularly more significantly in pigs that experienced VF. For instance, during occlusion, the dpca_w,tpe in pigs that developed VF increased from 4.92 ms to 44.51 ms, while in pigs that did not develop VF, it increased from 1.94 ms to only 7.52 ms.

Comparison of Index Behavior

The temporal variation trends of dpca_w,tpe and tpca_pe were similar, both showing significant increases within the first 5 minutes of coronary occlusion. In dynamic analyses over various time intervals, the dpca_w,tpe index demonstrated good distinguishability, with sensitivity and specificity of 90.0% and 75.0%, respectively, significantly superior to the sensitivity of 80.0% and specificity of 69.0% for the tpca_pe index.

VF Prediction Analysis

On the Receiver Operating Characteristic (ROC) curve, the dpca_w,tpe index exhibited better predictive capability for VF, with an area under the curve (AUC) of 0.85 and a hazard ratio of 12.5, whereas the tpca_pe index had an AUC of 0.79 and a hazard ratio of 5.5. This indicates that dpca_w,tpe has an advantage in predicting imminent VF events compared to tpca_pe.

Conclusions and Significance

The T-wave morphological index based on time warping, dpca_w,tpe, better captures ischemia-induced VR dispersion changes. Compared to tpca_pe, dpca_w,tpe showed higher predictive accuracy in predicting imminent VF following myocardial infarction. The results of this study suggest that further clinical exploratory studies should be conducted to validate its applicability and value in human populations.

Research Highlights

1. Novel Time Warping Technique: Compared to the traditional T-peak-to-T-end interval, the time warping-based morphological analysis method (dpca_w,tpe) excels in capturing ischemia-induced VR changes.
2. Higher Predictive Accuracy: dfca_w,tpe demonstrated higher sensitivity and specificity in predicting VF events, with a significantly higher hazard ratio than the traditional tpca_pe index.
3. Clinical Application Potential: The results of this study provide new ideas and methods for early prediction of VF risk in acute myocardial infarction patients in clinical settings.

This study enhances understanding of the potential of ECG analysis methods based on time warping technology, offering new directions and tools for future heart disease prevention and treatment.