HeartUnloadNet: A Weakly-Supervised Cycle-Consistent Graph Network for Predicting Unloaded Cardiac Geometry from Diastolic States

2025-07-24

AI Summary

This study addresses the critical challenge of noninvasively reconstructing the unloaded (zero-pressure) intrinsic cardiac geometry from clinical images—a foundational requirement for personalized biomechanical modeling. To overcome data scarcity and poor generalizability, we propose a cycle-consistent graph neural network framework that integrates graph attention mechanisms, weakly supervised learning, and physiological/biophysical priors. The method employs bidirectional shape mapping between end-diastolic image-derived geometries and unloaded myocardial configurations, enforced by cycle-consistency constraints, enabling end-to-end prediction. Evaluated on 20,700 synthetic cases, it achieves sub-millimeter accuracy (Dice Similarity Coefficient = 0.986), with inference time of only 0.02 seconds per case—over 10⁵× faster than conventional methods. Crucially, it markedly reduces reliance on labor-intensive annotated data while enhancing clinical deployability.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Computer Vision: Medical and Biological Imaging

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

The unloaded cardiac geometry (i.e., the state of the heart devoid of luminal pressure) serves as a valuable zero-stress and zero-strain reference and is critical for personalized biomechanical modeling of cardiac function, to understand both healthy and diseased physiology and to predict the effects of cardiac interventions. However, estimating the unloaded geometry from clinical images remains a challenging task. Traditional approaches rely on inverse finite element (FE) solvers that require iterative optimization and are computationally expensive. In this work, we introduce HeartUnloadNet, a deep learning framework that predicts the unloaded left ventricular (LV) shape directly from the end diastolic (ED) mesh while explicitly incorporating biophysical priors. The network accepts a mesh of arbitrary size along with physiological parameters such as ED pressure, myocardial stiffness scale, and fiber helix orientation, and outputs the corresponding unloaded mesh. It adopts a graph attention architecture and employs a cycle-consistency strategy to enable bidirectional (loading and unloading) prediction, allowing for partial self-supervision that improves accuracy and reduces the need for large training datasets. Trained and tested on 20,700 FE simulations across diverse LV geometries and physiological conditions, HeartUnloadNet achieves sub-millimeter accuracy, with an average DSC of 0.986 and HD of 0.083 cm, while reducing inference time to just 0.02 seconds per case, over 10^5 times faster and significantly more accurate than traditional inverse FE solvers. Ablation studies confirm the effectiveness of the architecture. Notably, the cycle-consistent design enables the model to maintain a DSC of 97% even with as few as 200 training samples. This work thus presents a scalable and accurate surrogate for inverse FE solvers, supporting real-time clinical applications in the future.

Problem

• Predict unloaded cardiac geometry from diastolic states
• Overcome computational expense of traditional FE solvers
• Enable real-time clinical applications with deep learning

Innovation

• Weakly-supervised cycle-consistent graph network
• Graph attention architecture with biophysical priors
• Bidirectional prediction enabling partial self-supervision