Vendor-Neutral Deep Learning Reconstruction Reduces T2 Liver MRI Scan Time and Enhances Image Quality

Vendor-neutral deep learning reconstruction (SwiftMR) shortened T2-weighted liver MRI by ~9s, improved sharpness and SNR (p<0.001), but motion artifacts increased; single breath-hold reduced spatial mismatch in 86 patients.

• Retrospective study of 86 patients undergoing 3-T liver MRI; two readers assessed routine two breath-hold T2WI versus accelerated SwiftMR protocols with one or two breath-holds.
• SwiftMR significantly improved sharpness and SNR (p<0.001) but increased motion artifacts; single breath-hold reduced spatial mismatch (p<0.05).
• Lesion conspicuity was not significantly different; limitation includes retrospective design and single-center data.

RadPigeon summaries are original and for information only. They are not clinical advice.