Technology Category
Computer Vision: Multi-modal VisionMachine Learning: Multimodal LearningIntelligent Robots: Multimodal Perception & Sensor Fusion
UniFS: Unified Multi-Contrast MRI Reconstruction via Frequency-Spatial Fusion
🤖 AI Summary
Current multi-contrast MRI reconstruction methods suffer from poor generalizability: they require separate training for each k-space undersampling pattern and neglect cross-contrast frequency-domain complementarity. To address this, we propose UniFS—the first unified reconstruction framework capable of adapting to diverse undersampling patterns without retraining. Its core innovations are: (1) cross-contrast frequency-domain fusion, explicitly modeling inter-contrast frequency synergies; (2) adaptive mask prompting, decoupling sampling-pattern priors from the reconstruction network; and (3) dual-branch complementary optimization, jointly modeling frequency- and spatial-domain features. Evaluated on BraTS and HCP datasets, UniFS achieves state-of-the-art performance across multiple acceleration factors and, critically, generalizes effectively to *unseen* undersampling patterns—demonstrating superior robustness and clinical applicability compared to prior methods.
Technology Category
Application Category
📝 Abstract
Recently, Multi-Contrast MR Reconstruction (MCMR) has emerged as a hot research topic that leverages high-quality auxiliary modalities to reconstruct undersampled target modalities of interest. However, existing methods often struggle to generalize across different k-space undersampling patterns, requiring the training of a separate model for each specific pattern, which limits their practical applicability. To address this challenge, we propose UniFS, a Unified Frequency-Spatial Fusion model designed to handle multiple k-space undersampling patterns for MCMR tasks without any need for retraining. UniFS integrates three key modules: a Cross-Modal Frequency Fusion module, an Adaptive Mask-Based Prompt Learning module, and a Dual-Branch Complementary Refinement module. These modules work together to extract domain-invariant features from diverse k-space undersampling patterns while dynamically adapt to their own variations. Another limitation of existing MCMR methods is their tendency to focus solely on spatial information while neglect frequency characteristics, or extract only shallow frequency features, thus failing to fully leverage complementary cross-modal frequency information. To relieve this issue, UniFS introduces an adaptive prompt-guided frequency fusion module for k-space learning, significantly enhancing the model's generalization performance. We evaluate our model on the BraTS and HCP datasets with various k-space undersampling patterns and acceleration factors, including previously unseen patterns, to comprehensively assess UniFS's generalizability. Experimental results across multiple scenarios demonstrate that UniFS achieves state-of-the-art performance. Our code is available at https://github.com/LIKP0/UniFS.
Problem
Research questions and friction points this paper is trying to address.
Generalizes across multiple k-space undersampling patterns without retraining
Integrates frequency and spatial information for cross-modal MRI reconstruction
Enhances generalization via adaptive prompt-guided frequency fusion module
Innovation
Methods, ideas, or system contributions that make the work stand out.
Unified model handles multiple undersampling patterns without retraining
Adaptive prompt-guided frequency fusion enhances generalization performance
Integrates cross-modal frequency, mask-based prompt, and dual-branch refinement modules
🔎 Similar Papers
No similar papers found.
