Q-space Guided Collaborative Attention Translation Network for Flexible Diffusion-Weighted Images Synthesis

2025-05-14

AI Summary

This work addresses the challenging synthesis of multi-shell high-angular-resolution diffusion imaging (MS-HARDI) data under flexible, non-uniform q-space sampling. We propose a generative framework that integrates structural MRI priors to enhance anatomical fidelity. Methodologically, we introduce a novel q-space-guided co-attention mechanism enabling robust modeling of arbitrary irregular q-sampling schemes; incorporate anatomical fidelity constraints and parameter-map consistency regularization to jointly encode diffusion-weighted signal characteristics, q-space geometry, and structural MRI anatomy; and employ multimodal feature fusion with structure-guided adversarial training. Evaluated on the Human Connectome Project (HCP) dataset, our approach achieves state-of-the-art performance—outperforming 1D-qDL, 2D-qDL, MESC-SD, and QGAN across diffusion parameter map estimation, fiber orientation reconstruction, and fine-grained structural preservation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Learning with Manifolds
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Web data generation and simulation
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems

Abstract

This study, we propose a novel Q-space Guided Collaborative Attention Translation Networks (Q-CATN) for multi-shell, high-angular resolution DWI (MS-HARDI) synthesis from flexible q-space sampling, leveraging the commonly acquired structural MRI data. Q-CATN employs a collaborative attention mechanism to effectively extract complementary information from multiple modalities and dynamically adjust its internal representations based on flexible q-space information, eliminating the need for fixed sampling schemes. Additionally, we introduce a range of task-specific constraints to preserve anatomical fidelity in DWI, enabling Q-CATN to accurately learn the intrinsic relationships between directional DWI signal distributions and q-space. Extensive experiments on the Human Connectome Project (HCP) dataset demonstrate that Q-CATN outperforms existing methods, including 1D-qDL, 2D-qDL, MESC-SD, and QGAN, in estimating parameter maps and fiber tracts both quantitatively and qualitatively, while preserving fine-grained details. Notably, its ability to accommodate flexible q-space sampling highlights its potential as a promising toolkit for clinical and research applications. Our code is available at https://github.com/Idea89560041/Q-CATN.

Problem

• Synthesize multi-shell high-angular resolution DWI from flexible q-space sampling
• Leverage structural MRI data for complementary information extraction
• Preserve anatomical fidelity and fine-grained details in DWI synthesis

Innovation

• Q-space guided collaborative attention mechanism
• Flexible q-space sampling adaptation
• Task-specific constraints for anatomical fidelity