Advancing 3T Vessel Wall Imaging through Combined Deep Learning Acceleration and Reconstruction for Rapid, High-Resolution Scans

Motivation: Achieving high-resolution vessel wall imaging (VWI) within a clinically feasible scan time remains challenge. Goal(s): We aimed to combine deep learning (DL) techniques to reduce scan times and improve image quality, enabling both rapid routine imaging and high-resolution VWI. Approach: A 3T MRI was used with DL acceleration and DL reconstruction to perform black blood T1-weighted imaging at 0.7 mm and 0.4 mm isotropic resolutions. Results: The combined DL techniques allowed for 0.7 mm resolution scans in under 2 minutes and 0.4 mm resolution scans in 7.5 minutes, achieving diagnostically acceptable quality and pushing resolution boundaries. Impact: This study demonstrates the feasibility of combining two DL techniques for rapid and high-resolution vessel wall imaging, offering unprecedented quality and speed at 3T. This approach also provides insights for applying DL methods in other challenging imaging sequences.