Optimizing Accelerated Protocols for Clinical T1w Vessel Wall Imaging: A Comparison of Deep Learning, Compressed Sensing, and CAIPI Methods

Motivation: Clinical intracranial VWI requires high spatial resolution and effective blood/CSF suppression, leading to long scan times, motion artifacts, and increased patient burden. Goal(s): To evaluate optimized protocols for intracranial VWI using accelerated 3D T1w SPACE sequences. Approach: We evaluated DL, CS, and CAIPI acceleration on 20- and 64-channel coils to reduce scan times, comparing them to the clinical GRAPPA protocol. Results: The 64-channel CS protocol with 6x acceleration matched clinical image quality and SNR with reduced scan time. CS and DL sequences had similar image quality and SNR on the 64-channel coil, with DL showing better noise suppression on the 20-channel coil. Impact: This study optimizes intracranial VWI using 3D T1w SPACE with DL, CS, and CAIPI for 20- and 64-channel coils, achieving reduced scan times. The 64-channel CS (6x) matched clinical image quality, while DL provided superior noise suppression on 20-channel coils.