Uncovering Arrhythmic Substrate in a Heart Failure Patient Using Digital Twins

Thaís de Jesus Soares1, Guilherme Martins Couto1, Yan B Werneck2, Daniel Keim Almeida1, Daniel moreira Leme1, João Pedro Banhato Pereira1, Filipe De Lima Namorato1, MATHEUS CARDOSO FAESY3, Tiago Dutra Franco4, Fabrício Santos4, Raul Pereira Barra4, Marcelle Cristina da Silva Bastos Vasconcelos5, THAIZ RUBERTI SCHMAL6, Thiago Goncalves Schroder e Souza7, Rafael Sachetto Oliveira8, Joventino de Oliveira Campos1, Rodrigo Weber dos Santos1
1Federal University of Juiz de Fora, 2PPGMC-UFJF, 3UFJF, 4Universidade Federal de Juiz de Fora, 5Department of Internal Medicine, Federal University of Juiz de Fora (UFJF), Juiz de Fora, Brazil, 6EBSERH University Hospital Juiz de Fora, 7University Hospital of the UFJF, 8Univerisidade Federal de S�o Jo�o del-Rei


Abstract

Systolic dysfunction and extensive non-ischemic myocardial fibrosis are common structural features in heart failure patients at high risk of ventricular arrhythmias and sudden cardiac death (SCD), a leading cause of mortality worldwide. In Brazil, these patients often undergo invasive electrophysiological studies (EPS), involving catheter insertion and programmed electrical stimulation to induce arrhythmias for diagnostic purposes. Despite its effectiveness, EPS is costly, invasive, and poses significant risks.

In this work, we propose the use of a personalized digital twin as a minimally invasive alternative for the virtual replication of EPS. Magnetic resonance imaging (MRI) and electrocardiogram (ECG) data were obtained from a heart failure patient treated at the university hospital of the Federal University of Juiz de Fora (HU-UFJF). Ventricular structures and fibrotic regions were segmented from the MRI images, allowing the construction of a three-dimensional computational cardiac model that incorporates tissue heterogeneity. The model differentiates three regions: healthy myocardium, fibrotic tissue, and a border zone (BZ). The healthy myocardium was modeled as a homogeneous conductive medium, while dense fibrosis was represented as a completely non-conductive region. The BZ was modeled in layers, with an increasing proportion of active cells from the fibrotic core to the healthy tissue.

Sixty electrophysiological variants of a single cardiac digital twin were generated, with variable BZ thicknesses and randomly generated microscopic fibrosis patterns in the border zone. Model conductivity was calibrated based on the patient's QRS complex duration from the ECG. The calibrated models were then used to virtually simulate the clinical EPS protocol.

The results showed that only fibrotic configurations with thick and heterogeneous border zones were able to reproduce the arrhythmias observed in the real patient, highlighting the critical role of microstructural variability in the formation of ventricular arrhythmias.