Transformer-Guided Graph Attention for Direct Cardiac Mesh Reconstruction: A Structural Digital Twin Framework
Researchers have developed a novel end-to-end network for direct cardiac mesh reconstruction from 3D medical images, bypassing traditional segmentation and mesh generation steps. This approach utilizes a 3D Swin Transformer for feature extraction and a Graph Attention Network (GAT) to deform a template mesh to the cardiac boundary. Tested on the MM-WHS 2017 benchmark, the method achieved competitive segmentation scores and improved mesh quality, producing simulation-ready meshes in a single forward pass. AI
IMPACT Streamlines the creation of patient-specific cardiac models, potentially accelerating clinical adoption of digital twin technology.