Super-resolution network with confidence maps boosts non-contrast coronary MRA quality

PABCNet super-resolves non-contrast coronary MRA, achieving a vessel ROI-PSNR of 32.68 dB (2.18 dB better than the second-best method) and providing pixel-wise confidence maps. Radiologists rated it superior in coronary clarity, cardiac structure depiction, and artifact suppress…

• PABCNet is a super-resolution pipeline for non-contrast coronary MRA that incorporates inter-slice phase alignment and Bayesian confidence quantification.
• In experiments, vessel-region ROI-PSNR was 32.68 dB, outperforming other methods by 2.18 dB, with a calibration error of 0.0234 for uncertainty maps.
• Radiologists rated PABCNet superior to both comparative methods and original high-resolution scans in coronary clarity, cardiac structure depiction, and artifact suppression.

Automated summary

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