End-to-end deep learning auto-navigation and reconstruction for accelerated free-breathing motion-resolved MRI

Motivation: State-of-the-art and clinical abdominal MRI techniques are still limited by motion, acquisition time, and reconstruction time. Goal(s): To develop an end-to-end deep learning approach including auto-navigation and motion-resolved reconstruction for fast and robust free-breathing T1-weighted MRI with a scan time of only 1 minute and reconstruction times of <1 minute. Approach: Acquisition used a 3D T1-weighted golden-angle stack-of-stars pulse sequence with RANGR auto-navigation and Movienet reconstruction for motion-resolved imaging implemented on a clinical scanner. Two expert radiologists evaluated image quality. Results: The proposed deep learning technique enables robust free-breathing MRI with a 1-minute scan time that compares favorably to the clinical standard. Impact: The combination of deep learning auto-navigation and motion-resolved reconstruction enables fast and robust free-breathing abdominal MRI, which has the potential to reduce the number of repeat scans and increase efficiency compared to current clinical standards.