Spatial-Temporal Mercer Kernel for Premature Ventricular Complex Estimation in Electrocardiographic Imaging


Abstract:

Cardiovascular diseases are a leading cause of death globally, highlighting the need for accurate, noninvasive diagnostic tools. Electrocardiographic imaging (ECGI) has emerged as a promising technique for reconstructing epicardial potentials from body surface measurements, offering an alternative to invasive procedures. This work presents a novel spatio-temporal separable Mercer kernel framework for estimating premature ventricular complex (PVC) origins, aiming to enhance the precision of ECGI-based localization. We incorporate spatial-temporal correlations in the kernel and use ADAM optimization to model epicardial potential dynamics. The proposed approach is evaluated using real ECGI data, and its performance is compared to traditional methods, including the Tikhonov regularization. Results show that the Gaussian kernel achieves the lowest NSME (0.3439) and smallest localization error (12.55 mm), outperforming the Laplacian kernel (NSME = 0.3503; 23.99 mm) and Tikhonov (NSME = 0.5036; 36.48 mm). This study contributes to advancing ECGI as a reliable tool for guiding catheter ablation in PVC patients, improving clinical outcomes through previous localization.

Año de publicación:

2025

Keywords:

    Fuente:

    scopusscopus

    Tipo de documento:

    Other

    Estado:

    Acceso restringido

    Áreas de conocimiento:

    • Enfermedad cardiovascular
    • Aprendizaje automático

    Áreas temáticas de Dewey:

    • Enfermedades
    • Métodos informáticos especiales
    • Análisis
    Procesado con IAProcesado con IA

    Objetivos de Desarrollo Sostenible:

    • ODS 3: Salud y bienestar
    • ODS 17: Alianzas para lograr los objetivos
    • ODS 9: Industria, innovación e infraestructura
    Procesado con IAProcesado con IA