Deep learning reconstruction accelerates 3D knee bone MRI from 30 min to 2.5–10 min

Deep learning reconstruction (DnCNN) accelerated 3D knee bone MRI from 30 minutes to as low as 2.5 minutes in a feasibility study, with iterative reconstruction reducing denoising artifacts at 5-minute scans.

• The DnCNN alone yielded the most favorable quantitative metrics in healthy volunteers.
• Iterative FISTA-based reconstruction helped mitigate denoising-induced streaking artifacts, especially for the 5-minute scan subset.
• In a single pathological case, AI-reconstructed MRI visualized a tibial plateau fracture but with less detail than photon-counting CT.

Automated summary

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