Dynamic-Aware Spatio-temporal Representation Learning for Dynamic MRI Reconstruction

2025-01-15

AI Summary

To address poor reconstruction quality, low optimization efficiency, and hyperparameter sensitivity in dynamic MRI undersampled reconstruction, this paper proposes Dynamic-Aware Implicit Neural Representations (DA-INR). DA-INR is the first method to explicitly embed dynamic priors into the INR architecture, jointly modeling spatiotemporal continuity and temporal redundancy in the image domain for fully unsupervised reconstruction. By integrating coordinate encoding with explicit temporal redundancy modeling, DA-INR achieves both high representational capacity and rapid convergence—effectively alleviating the slow optimization and tedious hyperparameter tuning that plague conventional INRs in dynamic medical imaging. Experiments demonstrate that under extreme undersampling, DA-INR achieves significantly higher PSNR and SSIM than state-of-the-art methods; single-reconstruction optimization time is reduced by over 60%; hyperparameter robustness is substantially improved; and no ground-truth labels are required.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Dynamic MRI reconstruction, one of inverse problems, has seen a surge by the use of deep learning techniques. Especially, the practical difficulty of obtaining ground truth data has led to the emergence of unsupervised learning approaches. A recent promising method among them is implicit neural representation (INR), which defines the data as a continuous function that maps coordinate values to the corresponding signal values. This allows for filling in missing information only with incomplete measurements and solving the inverse problem effectively. Nevertheless, previous works incorporating this method have faced drawbacks such as long optimization time and the need for extensive hyperparameter tuning. To address these issues, we propose Dynamic-Aware INR (DA-INR), an INR-based model for dynamic MRI reconstruction that captures the spatial and temporal continuity of dynamic MRI data in the image domain and explicitly incorporates the temporal redundancy of the data into the model structure. As a result, DA-INR outperforms other models in reconstruction quality even at extreme undersampling ratios while significantly reducing optimization time and requiring minimal hyperparameter tuning.

Problem

• Dynamic MRI Reconstruction
• Data缺失 (Data Missing)
• Image Quality Optimization

Innovation

• DA-INR
• Dynamic MRI Reconstruction
• Temporal-Spatial Information