Physics in medicine and biology, 20 3 2025 Analysis and generation of the fibrosis textures based on histology data of the human hearts with non-ischemic cardiomyopathy. Okenov A, Nezlobinsky T, Glashan CA, Zeppenfeld K, Vandersickel N, Panfilov A
Objective
Cardiac arrhythmias, marked by irregular and/or fast heartbeats, can severely compromise heart function and pose serious health risks. Fibrosis alters the propagation of electrical signals in the heart and plays a critical role in arrhythmogenesis. The impact of fibrosis depends not only on its amount, but also on its geometric features and spatial distribution. Therefore, understanding and replicating the structural characteristics of fibrotic tissue is essential for accurate arrhythmia modeling, particularly in computational and AI-based studies. Approach: We analyzed a dataset of histological images from ten patients with non-ischemic cardiomyopathy and sustained ventricular tachycardia. A statistical evaluation was performed on key geometric descriptors of fibrosis---size, elongation, orientation, and solidity---all of which influence the pro-arrhythmic potential of the tissue. To address the challenge of generating realistic fibrotic textures, we proposed the Direct Sampling method. This geostatistical image synthesis technique was used to produce large-scale synthetic fibrosis textures based on real data. Main Results: Fibrotic cluster size spans six orders of magnitude within individual hearts. Fibrosis is generally elongated, with median elongation values between 2.19 and 2.63 depending on the patient and the myocardial layer. It has preferential alignment within the myocardial wall, especially in the midwall layers. The solidity of the clusters decreases nearly logarithmically with increasing fibrosis size. The Direct Sampling method successfully reproduces the complexity and heterogeneity of real fibrotic patterns and generates diverse structurally realistic samples from a single input texture. Significance: This study provides a comprehensive quantitative characterization of the geometric variability of fibrosis in non-ischemic cardiomyopathy and demonstrates the effectiveness of the Direct Sampling method in generating realistic non-ischemic fibrotic patterns. This work contributes to the development of more realistic heart models, which can enhance computational studies of arrhythmias.
