TensorLDM: A Component-Wise Latent Diffusion Model for Volumetric DTI Reconstruction from Sparse DWIs

2026-06-24

AI Summary

This work addresses the challenge of reconstructing diffusion tensors from sparse diffusion-weighted images (DWIs), a task in which existing deep learning methods often yield anatomically inconsistent or physically implausible results. To overcome this, the authors propose a component-wise latent diffusion model that employs a dual-branch encoder to separately process the diagonal and off-diagonal elements of the diffusion tensor. The architecture integrates an anatomy-conditioned autoencoder, cross-component attention mechanisms, and a mixture-of-experts DWI conditioner to model inter-component dependencies under shared DWI constraints, thereby enforcing both anatomical consistency and physical plausibility. Evaluated on the Human Connectome Project (HCP) dataset, the model achieves state-of-the-art or comparable voxel-wise reconstruction accuracy with a remarkably low symmetric positive-definite (SPD) violation rate of 1.54%, significantly enhancing the fidelity of downstream fiber tractography.

Abstract

Reconstructing diffusion tensors from sparse DWIs is critical for accelerating Diffusion Tensor Imaging (DTI) in clinical settings, yet current deep learning approaches frequently yield anatomically inconsistent or physically implausible tensors. We introduce TensorLDM, a component-wise latent diffusion model that processes the six tensor components through two group-specific encoders (for diagonal and off-diagonal elements) while maintaining anatomical consistency via shared DWI conditioning. TensorLDM uses an Anatomy-Conditioned Autoencoder that encourages the latent to focus on tensor properties rather than re-encoding structural information. A shared Cross-Component Attention (CCA) mechanism, applied in both autoencoder refinement and diffusion fine-tuning, models inter-component dependencies, while a Mixture-of-Experts (MoE) DWI conditioner provides component-adaptive conditioning. On the Human Connectome Project (HCP) dataset under a single-shell, four-volume sparse acquisition, TensorLDM produces the most accurate downstream tractography and tensors with near-ground-truth physical validity (SPD-violation rate 1.54% vs. 1.40%), with the best or comparable voxel-wise reconstruction accuracy. Geodesic tensor error measured by the Log-Euclidean Metric (LEM) corroborates these gains.

Problem

• Diffusion Tensor Imaging
• Sparse DWIs
• Tensor Reconstruction
• Anatomical Consistency
• Physical Plausibility

Innovation

• Latent Diffusion Model
• Diffusion Tensor Imaging
• Cross-Component Attention
• Mixture-of-Experts
• Anatomy-Conditioned Autoencoder