Accelerated 1.2-minute 4-echo UTE-based CT-like Imaging using CG-SENSE and Deep Learning-based Denoising Reconstruction (DLR)

Motivation: Multi-echo UTE provides CT-like image contrast, which allows bone assessment without radiation. However, long scan-time and low resolution may hinder it wider clinical adoption. Goal(s): We aimed to reduce scan-time and/or improve image resolution/sharpness of mecho-UTE using a CG-SENSE and Deep Learning-based Denoising Reconstruction. Approach: On patients, 5.8min 0.8 mm3 isotropic 3D 4-echo UTE sequence was acquired and 2-5X retrospectively accelerated. The resolution and sharpness of CT-like images were measured and compared for gridding (GRID) and CG-SENSE+DLR. Results: DLR+CG-SENSE enables up to 5X acceleration (1.2min) with comparable resolution/sharpness compared to the reference 5.8min acquisition. Impact: CG-SENSE+DLR improves the resolution and sharpness of CT-like images that couples with a potential 1.2min scan time would encourage wider clinical adoption of radiation-free UTE for bone assessment.