Hybrid learning: a combination of self-supervised and supervised learning for joint MRI reconstruction and denoising in low-field MRI.
Authors
Affiliations (4)
Affiliations (4)
- Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University Grossman School of Medicine, 227 E 30th St, New York, 10016, United States.
- Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, New York University Grossman School of Medicine, 227 E 30th St, New York, New York, 10016, United States.
- Biomedical Engineering and Imaging Institute and Department of Radiology, Icahn School of Medicine at Mount Sinai, 1470 Madison Ave, New York, New York, 10029-5674, United States.
- Electrical and Computer Engineering Department, New York University Tandon School of Engineering, 370 Jay Street, Brooklyn, New York, New York, 11201, United States.
Abstract
Deep learning has demonstrated strong potential for MRI reconstruction. However, conventional supervised learning requires high-quality, high-SNR reference data for network training, which are often difficult or impossible to obtain, particularly in low-field MRI. Self-supervised learning eliminates the need for reference training data but may suffer from degraded performance under low-SNR conditions. To address these limitations, we propose hybrid learning, a new training framework that integrates self-supervised and supervised learning for joint MRI reconstruction and denoising when only low-SNR training data are available.

Methods: Hybrid learning is implemented in two sequential stages. In the first stage, self-supervised learning is applied to fully sampled low-SNR data to generate higher-quality pseudo-references. In the second stage, these pseudo-references are then used as targets for supervised learning to reconstruct and denoise undersampled noisy data. The proposed method was evaluated in four experiments using simulated and real noisy MRI data of the breast, lung and brain across different field strengths (0.3T to 3T), sampling trajectories (Cartesian, spiral, and radial), noise levels, and undersampling ratios. 

Results: Hybrid learning consistently improved reconstruction quality relative to both supervised and self-supervised baselines under different acceleration rates, noise levels, and sampling patterns in all experiments. Compared with standard supervised learning using noisy references, it achieved up to 167.70% higher structural similarity index (SSIM), 95.41% lower normalized mean square error (NMSE), and 90.70% lower high-frequency error norm (HFEN). Compared with standard self-supervised learning, it achieved up to 23.88% higher SSIM, 60.85% lower NMSE, and 49.13% lower HFEN. 

Conclusion: Hybrid learning enables improved MRI reconstruction under low-SNR imaging conditions by jointly addressing noise and undersampling. It provides a practical solution for robust deep learning-based reconstruction and is particularly well suited for applications such as low-field MRI, where image quality is limited by reduced SNR.