Compact model improves 3T to 7T MRI image synthesis with fewer resources
DGCM-SSFA: Dual-Granularity Cross-Modulation and Sparse Strip Frequency Attention for Efficient 3T-to-7T MRI Synthesis
Computer Vision and Pattern Recognition
Summary
Getting super-clear brain images usually needs expensive, high-powered MRI machines. This paper’s authors designed a smaller deep learning model that turns standard MRI scans into ultra-high-quality ones like those from powerful machines, but with much less computing needed. Their model uses smart techniques to focus on important image details and large-scale patterns, making it efficient and accurate. Tests show it nearly matches the quality of much bigger models while being easier to use in clinics with limited resources.
What this means in practice
- •For clinical radiology teams: Generate high-quality 7T-like MRI images from routine 3T scans to improve diagnostic detail without extra scanning costs.
- •For medical device developers: Build efficient MRI image enhancement software that runs well on limited hardware by using compact multi-domain attention models.$Commercial implications: Enables production of affordable MRI enhancement tools for hospitals by reducing computational resources while retaining image quality.
Authors
Zhichun Fan, Zhiqiang Chen, Lanling Liu, Lingjun Liu, Zhonghua Xie
Abstract
Synthesizing high-fidelity 7T MRI from routinely acquired 3T scans offers a practical pathway to enhanced neuroimaging without the prohibitive cost of ultra-high-field scanners. While existing deep learning approaches have achieved promising results, state-of-the-art methods typically rely on either large-scale architectures exceeding 45 million parameters or computationally intensive ensemble strategies, limiting their practicality in resource-constrained clinical environments. In this work, we propose a compact multi-domain attention framework, termed DGCM-SSFA, that achieves competitive synthesis quality with only 11M parameters. The proposed architecture integrates two complementary modules into a U-Net Transformer backbone. First, a Dual-Granularity Cross-Modulated Feature Modulator (DGCM) is embedded at multiple hierarchical levels to enhance local edge and texture details through Sobel-guided spatial attention and depthwise frequency-aware channel attention with cross-branch gating. Second, a Sparse Strip Frequency Attention (SSFA) module at the bottleneck captures global spatial context via strip-pooled attention with near-linear complexity, while simultaneously performing FFT-based spectral modulation in the frequency domain. These two modules operate synergistically with the backbone's native channel attention, forming a triple-granularity feature refinement pipeline. Extensive experiments on the UNC and BNU 3T-to-7T MRI datasets across T1w and T2w modalities demonstrate that DGCM-SSFA achieves strong structural fidelity, attaining an SSIM of 0.8431 on BNU T2w---approaching the best benchmark value of 0.8624---and 0.8100 on BNU T1w, while using substantially fewer parameters than Transformer-based alternatives such as ResViT (123.4M) and FS-RWKV (45.4M).