AI-accelerated 3D vessel wall MRI cut scan time 42% with noninferior image quality and lesion assessment

In 30 patients, deep learning-accelerated 3D vessel wall MRI cut scan time 42% (9:15 to 5:20) and was noninferior to conventional MRI for all image-quality parameters, with perfect lesion-detection agreement (kappa=1.00).

• 79 vessel wall lesions were identified; inter-protocol agreement was perfect for lesion detection and wall-thickening pattern (kappa=1.00; 58 eccentric, 21 concentric).
• Contrast enhancement agreement was excellent (kappa=0.89), and lumen-background contrast-to-noise ratio was comparable despite 10-fold acceleration (P=0.95).
• Sonic DL improved cerebrospinal fluid suppression (P=0.002); inter-reader agreement was good to excellent (kappa=0.79-0.88).

Automated summary

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