CineMesh4D Reconstructs 4D Whole Heart from Sparse MRI Slices
New AI pipeline reconstructs beating 3D heart from standard MRI slices in real-time.
Accurate 3D+t (4D) whole-heart mesh reconstruction from cine MRI is clinically vital but technically challenging due to sparse 2D slice sampling and the coupling of shape and motion. Current approaches often reconstruct only subsets of chambers or a single cardiac phase. Researchers from the National University of Singapore and other institutions have developed CineMesh4D, a novel end-to-end pipeline that directly reconstructs a personalized whole-heart mesh from multi-view 2D cine MRI via cross-domain mapping. The method uses a differentiable rendering loss to supervise the 3D+t mesh from multi-view sparse contours, bypassing the need for dense 3D ground truth.
Additionally, CineMesh4D introduces a dual-context temporal block that fuses global and local cardiac temporal information to capture high-dimensional sequential patterns of the beating heart. In quantitative and qualitative evaluations, CineMesh4D outperforms existing approaches in reconstruction quality and motion consistency. This provides a practical pathway for personalized real-time cardiac assessment, with potential to reduce scan time and improve diagnostic accuracy. The code will be publicly released once the manuscript is accepted, making this approach accessible for further research and clinical translation.
- End-to-end 4D pipeline reconstructs whole-heart mesh from multi-view 2D cine MRI slices
- Differentiable rendering loss enables supervision from sparse 2D contours without 3D ground truth
- Dual-context temporal block fuses global and local cardiac motion patterns for better consistency
Why It Matters
Enables personalized real-time cardiac assessment from standard MRI without contrast or long scans.