Towards 3D heart mesh generation using contactless radar imaging and physics-informed neural network

2026-05-25

AI Summary

Reconstructing high-fidelity, topologically continuous 3D cardiac meshes from millimeter-wave synthetic aperture radar (SAR) images is highly challenging: conventional approaches yield only sparse point clouds, while existing optical reconstruction networks struggle with the strong speckle noise and blurred boundaries inherent in SAR imagery. To address this, we propose SAR2Mesh, a framework that initializes from an anatomical template and employs a coarse-to-fine mesh deformation strategy, integrating geometry-aware 3D-to-2D feature projection with an end-to-end differentiable, physics-informed radar consistency loss to produce surfaces consistent with the original radar echoes. We introduce Cardiac Mesh-SAR, the first large-scale paired SAR–cardiac mesh dataset, and establish a novel paradigm combining topology preservation, multi-view feature sampling, and physical constraints. Experiments demonstrate that SAR2Mesh significantly outperforms existing image-driven methods, achieving high-precision, physically consistent 3D cardiac reconstructions while preserving anatomical fidelity.

Abstract

Cardiac function evaluation necessitates continuous, non-invasive monitoring, a capability limited in MRI. Millimeter-wave (mmWave) radar and its Synthetic Aperture Radar (SAR) mode offer a privacy-preserving and portable point-of-care clinical applications. However, reconstructing high-fidelity 3D cardiac geometry from SAR remains an open challenge. Traditional radar methods generate sparse point clouds that lack continuous surface topology. Meanwhile, direct application of optical reconstruction networks performs poorly due to the severe speckle noise and ambiguous boundaries inherent in SAR images. To bridge this gap, we propose SAR2Mesh, a novel framework that reformulates the task as a coarse-to-fine mesh deformation process. By initializing with a topological template, our approach explicitly preserves anatomical connectivity through progressive mesh deformation.We introduce a geometry-aware feature projection module to extract multi-view features via 3D-to-2D sampling, and a physics-informed radar loss to enforce consistency between the predicted geometry and raw radar echoes. Furthermore, we present Cardiac Mesh-SAR, the first large-scale paired SAR-mesh dataset. Extensive experiments demonstrate that SAR2Mesh significantly outperforms existing image-based baselines, achieving accurate and physically consistent cardiac reconstructions.

Problem

• 3D heart mesh generation
• contactless radar imaging
• SAR image reconstruction
• cardiac geometry
• non-invasive monitoring

Innovation

• SAR2Mesh
• physics-informed neural network
• 3D cardiac mesh reconstruction
• mmWave radar imaging
• geometry-aware feature projection