CaMVIA-3D: Cardiac Microarchitecture Volumetric Imaging and Analysis Pipeline in 3D

All articles of this category (opens in new window) Background: In response to pathogenic stimuli, the heart undergoes remodeling at the level of individual cardiomyocytes (CM), marked by changes in cell geometry, nucleation, and expansion of extracellular volume (ECV). These microarchitectural changes are central to the pathophysiology of diverse structural and genetic heart diseases. However, conventional 2D histology fails to capture true 3D cellular and extracellular morphology. To overcome this limitation and better understand cardiac microarchitecture in health and disease, we developed an imaging and deep learning pipeline to quantify 3D CM morphology and ECV. Methods: We developed CaMVIA-3D, a pipeline for 3D analysis of CM morphology and ECV. Cardiac tissues were fixed and sectioned at 150 µm. Sections were stained for sarcomeres (α-actinin), nuclei (DAPI), and WGA to label CM boundaries and extracellular space. High-resolution confocal z-stacks (100–150 µm in depth) were acquired from 8 to 10 regions of interest per section and preprocessed for background correction. CMs and nuclei were segmented in 3D using a custom-trained Omnipose model. Volume, eccentricity (length-to-width ratio), nuclei number, and ECV were automatically quantified. The model was trained on approximately 5,000 manually annotated CMs from control and HCM samples, including fibrotic and disarrayed regions. Results: To date, CaMVIA-3D has been applied to myocardial tissue from over 100 subjects to analyze the 3D morphology of more than 100,000 CMs. The method showed reproducible single-cell segmentations across subjects and independently processed tissue blocks. It enabled the detection of regional and disease-related differences in CM volume, eccentricity, nuclei number, and ECV. Quantification remained robust in fibrotic and disarrayed tissue. Segmentation accuracy was validated against approximately 900 manually curated 3D ground truth labels (IoU 94.5%). Compared to Cellpose-3D, CaMVIA-3D demonstrated significantly higher segmentation accuracy and markedly lower over-segmentation and under-segmentation error rates across species and genotypes. Conclusion: CaMVIA-3D is the first pipeline to enable high-throughput 3D analysis of cardiac microarchitecture at single-cell resolution in intact tissue. It enhances our understanding of structural remodeling in health and disease and offers a broadly applicable platform for improved patient stratification, genotype-specific diagnostics, and assessment of therapeutic effects on CM structure and extracellular remodeling. Publication History Article published online: 17 February 2026 © 2026. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany