Clinical value of a 10-minute deep-learning reconstruction fast spinecho knee magnetic resonance imaging protocol including T2 mapping

Motivation: T2 mapping is difficult to be implemented in clinics for long scan time. Goal(s): To propose a 10-minute fast spin echo knee MRI protocol including quantitative T2 map and morphologic images in identifying cartilage injury or grading cartilage degeneration. Approach: To subjectively and objectively assess the clinical value of knee imaging with different acceleration factor of 2 and 3 as well as using conventional and deep learning reconstruction algorithms (CR, DLR). Results: Both FSEDLR and DL-based quantitative T2 map with either an acceleration factor of 2 or 3 elevated diagnostic accuracy using a senior rater as a final diagnosis. Impact: A knee protocol including deep-learning based FSE structure imaging and quantitative T2 map could be facilitated in clinical application further elevating the diagnostic accuracy and performance.