Feasibility and Clinical Utility of Deep Learning-Based Composite Super-Resolution Reconstruction for Knee MRI

Motivation: The knee joint is prone to degenerative changes and injuries. Conventional knee MRI is important but often has low efficiency and reduced image quality. Deep learning-based super-resolution techniques show potential to address these issues, but clinical validation is needed to confirm practical benefits. Goal(s): Investigate the clinical benefits of a deep learning-based composite super-resolution reconstruction approach for knee MRI. Approach: 110 patients with each scanning conventional and composite super-resolution reconstruction knee MRI were included. Image quality was compared using objective and subjective assessments. Results: The composite super-resolution reconstruction significantly improved image quality, efficiency, and diagnostic value compared to conventional knee MRI. Impact: The practice of deep learning-based super-resolution reconstruction would be beneficial for knee MRI overall diagnostic quality and efficiency, further enhancing patient comfort and clinical workflow.