Advanced ZTE MR Lung Imaging: A Deep Learning Approach to Enhance SNR and Reduce Artifacts

Motivation: Zero Echo-Time (ZTE) MRI enables imaging of short T2 species, making it a preferred option for imaging lung parenchyma. In accelerated clinical protocols, small and diffuse structures can be obscured by noise, ringing or chemical shift artifacts. Goal(s): To evaluate the feasibility and utility of a deep learning (DL)-ZTE reconstruction for improving signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and reducing ringing and chemical shift artifacts in ZTE lung images. Approach: Testing the DL-ZTE model in ZTE lung images of healthy volunteers and Interstitial Lung Disease patients. Results: DL-ZTE images preserved SSIM, removed artifacts and significantly improved SNR and CNR metrics in the lung. Impact: The DL-ZTE model significantly enhances lung MRI quality by reducing artifacts and improving SNR while maintaining anatomical accuracy, facilitating more accurate and reliable clinical assessments of parenchymal abnormalities in ZTE lung imaging applications.