Bi-PT: Bidirectional Cross-Attention Point Transformers for Four-Chamber Heart Reconstruction from Sparse Cardiac MRI Data

2026-07-07

AI Summary

This work addresses the challenges of large reconstruction errors and inaccurate correspondences in 3D four-chamber cardiac mesh reconstruction from clinically acquired sparsely sampled MRI data. The authors propose a novel method that integrates bidirectional point-wise cross-attention with neural ordinary differential equations (NODEs). By enabling robust feature interaction between an anatomical atlas and sparse point clouds, and incorporating point-wise semantic label–guided Chamfer distance along with smoothness regularization, the approach achieves high-fidelity deformation modeling under local affine diffeomorphic constraints for the first time. Experimental results on real clinical datasets demonstrate that the proposed method significantly outperforms existing baselines, yielding accurate and robust reconstructions of the four-chamber heart geometry.

Abstract

We propose Bi-PT, a pipeline for reconstructing 3D four-chamber human heart meshes from clinical sparsely sampled cardiac magnetic resonance imaging (CMR) data. This work addresses the error-prone generation of 3D cardiac shape from a sparse point cloud (SPC) extracted from 2D long-axis and short-axis views used in routine clinical CMR protocols. Bi-PT enables accurate inference of the four-chamber heart mesh from the SPC by learning robust point features via bidirectional point cross-attention between an atlas and the SPC, together with per-point semantic labels that improve correspondence estimation. We formulate the deformation field as a Neural Ordinary Differential Equation (NODE) parameterized by a per-point affine transformation and translation to deform the atlas toward the target heart shape. By learning such a NODE, we can guarantee the deformation field to be a locally affine diffeomorphic deformation. We also integrate a semantic label loss into the Chamfer distance to encourage label-consistent correspondences and add a smoothness regularization to stabilize and improve the learning of the deformation field. Extensive experiments demonstrate that Bi-PT achieves accurate and robust performance compared to baselines.

Problem

• four-chamber heart reconstruction
• sparse cardiac MRI
• 3D shape generation
• point cloud
• cardiac mesh

Innovation

• Bidirectional Cross-Attention
• Neural ODE
• Sparse Point Cloud
• Diffeomorphic Deformation
• Semantic Label Loss