![[2602.20306] Shape-informed cardiac mechanics surrogates in data-scarce regimes via geometric encoding and generative augmentation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.20306] Shape-informed cardiac mechanics surrogates in data-scarce regimes via geometric encoding and generative augmentation
Summary
This article presents a novel framework for creating shape-informed surrogates in cardiac mechanics, enhancing predictions in data-scarce environments through geometric encoding and generative augmentation.
Why It Matters
The research addresses the challenge of accurately modeling cardiac mechanics in clinical settings where data is limited. By improving surrogate modeling techniques, this work has the potential to enhance patient-specific simulations, leading to better clinical decision-making and patient outcomes.
Key Takeaways
- Introduces a two-step framework for cardiac mechanics modeling.
- Decouples geometric representation from physics response for better generalization.
- Utilizes synthetic geometry generation for data augmentation.
- Compares PCA-based and DeepSDF-based strategies for encoding geometric variability.
- Demonstrates robustness to noisy and sparse data inputs.
Computer Science > Machine Learning arXiv:2602.20306 (cs) [Submitted on 23 Feb 2026] Title:Shape-informed cardiac mechanics surrogates in data-scarce regimes via geometric encoding and generative augmentation Authors:Davide Carrara, Marc Hirschvogel, Francesca Bonizzoni, Stefano Pagani, Simone Pezzuto, Francesco Regazzoni View a PDF of the paper titled Shape-informed cardiac mechanics surrogates in data-scarce regimes via geometric encoding and generative augmentation, by Davide Carrara and 5 other authors View PDF HTML (experimental) Abstract:High-fidelity computational models of cardiac mechanics provide mechanistic insight into the heart function but are computationally prohibitive for routine clinical use. Surrogate models can accelerate simulations, but generalization across diverse anatomies is challenging, particularly in data-scarce settings. We propose a two-step framework that decouples geometric representation from learning the physics response, to enable shape-informed surrogate modeling under data-scarce conditions. First, a shape model learns a compact latent representation of left ventricular geometries. The learned latent space effectively encodes anatomies and enables synthetic geometries generation for data augmentation. Second, a neural field-based surrogate model, conditioned on this geometric encoding, is trained to predict ventricular displacement under external loading. The proposed architecture performs positional encoding by using universal ventric...
