Modeling Myocardial Fibrosis Improves ECG-Based Inference in Dilated Cardiomyopathy

Lucas Arantes Berg1, Filipe De Lima Namorato1, Joventino de Oliveira Campos1, Bernardo Martins Rocha2, José Geraldo Soares de Aurora1, João Victor Granatto de Carvalho1, Thaís de Jesus Soares1, MATHEUS CARDOSO FAESY3, Thiago Goncalves Schroder e Souza4, THAIZ RUBERTI SCHMAL5, Rodrigo Weber dos Santos1
1Federal University of Juiz de Fora, 2Universidade Federal de Juiz de Fora, 3UFJF, 4University Hospital of the UFJF, 5EBSERH University Hospital Juiz de Fora


Abstract

Myocardial fibrosis is a key structural feature of dilated cardiomyopathy (DCM) and plays a critical role in altering ventricular activation patterns. However, its impact on reduced-order electrophysiological models used for ECG fitting remains insufficiently explored. In this work, we investigate the effect of incorporating fibrosis into a reduced-order Eikonal-based inference model for approximating the QRS complex in a patient with idiopathic DCM. Patient-specific fibrotic regions were derived from late gadolinium-enhanced magnetic resonance imaging (LGE-MRI) and incorporated into the computational model after being manually segmented by a clinician. The QRS complex was approximated under two modeling assumptions: (i) without explicit representation of fibrosis, using a globally reduced conduc-tion velocity, and (ii) with spatially resolved fibrotic regions affecting condu-ction. Preliminary results show that, without modeling fibrosis, the QRS complex was approximated with a mean Pearson correlation coefficient (PCC) of 0.87 by reducing the reference conduction velocity by 80%. When fibrotic regions were incorporated with a slow conduction velocity close to zero, the model achieved a PCC of 0.90, indicating improved agreement with clinical electrocardiography. These findings suggest that explicitly modeling fibrosis provides a more physiologically consistent representation of condu-ction abnormalities in DCM and improves ECG-based inference, supporting its inclusion in reduced-order cardiac digital twin models.