Transformer-Guided Graph Attention for Direct Cardiac Mesh Reconstruction: A Structural Digital Twin Framework

2026-06-11

AI Summary

This work proposes the first end-to-end deep learning framework that directly generates smooth, simulation-ready cardiac surface meshes from 3D medical images, eliminating the need for separate segmentation, Marching Cubes reconstruction, and manual post-processing. By unifying segmentation and mesh generation into a single forward pass, the method achieves clinical deployability and computational efficiency. It employs a 3D Swin Transformer encoder–decoder to extract volumetric features and refines an initial template mesh through a graph attention network that iteratively deforms it to accurately conform to cardiac boundaries, ensuring both geometric fidelity and topological correctness. Evaluated on the MM-WHS 2017 dataset, the approach attains Dice scores of 0.84 and 0.83 for CT and MRI, respectively, with a mean Chamfer distance of 1.8 mm and 95% Hausdorff surface distances under 5 mm, producing high-quality meshes in a single inference.

Abstract

Building patient-specific cardiac models sits at the heart of precision cardiology, yet getting those models into clinical use keeps running into the same wall: mesh generation is slow, messy, and frustrating. The standard workflow -- segmenting the image, running Marching Cubes, and then manually cleaning up the result -- is time-consuming, inconsistent across operators, and demands specialist knowledge most clinical teams do not have. We take a fundamentally different approach. Instead of treating segmentation and mesh generation as two separate problems, we train a single end-to-end network that goes directly from a raw 3D medical image to a smooth, simulation-ready cardiac surface mesh. The core is a 3D Swin Transformer encoder-decoder that extracts volumetric features from CT or MRI volumes, paired with a Graph Attention Network (GAT) head that iteratively deforms a template mesh to fit the patient's cardiac boundary. We tested on the MM-WHS 2017 benchmark using both CT and MRI. Segmentation scores were competitive (Dice of 0.84 on CT, 0.83 on MRI), but the primary focus is mesh quality: mean Chamfer distance of 1.8 mm, with 95th-percentile surface distance below 5 mm. Every mesh is produced in a single forward pass -- no Marching Cubes, no smoothing filters, no manual cleanup. We argue that for cardiac digital twin pipelines, geometric fidelity and topological correctness matter more than pixel-level Dice scores. By removing the post-processing bottleneck, this approach makes patient-specific cardiac simulation substantially more accessible for clinical use.

Problem

• cardiac mesh reconstruction
• patient-specific modeling
• clinical translation
• mesh generation bottleneck
• digital twin

Innovation

• Transformer-Guided Graph Attention
• Direct Mesh Reconstruction
• Cardiac Digital Twin
• End-to-End Mesh Generation
• 3D Swin Transformer