Super-resolution deep learning-enhanced 2D MRCP improves main pancreatic duct visualization compared with 3D MRCP.

Area of Science:

• Radiology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• The comparative effectiveness of 2D and 3D MRCP for visualizing the main pancreatic duct (MPD) is not fully understood.
• The impact of super-resolution deep learning reconstruction (SR-DLR) on MPD visualization requires further investigation.

Purpose of the Study:

• To compare the image quality and MPD visualization of 3D MRCP with SR-DLR (3D-SR), conventional 2D MRCP with compressed sensing (2D-CS), and 2D MRCP with SR-DLR (2D-SR).
• To explore the diagnostic performance of these techniques for detecting MPD abnormalities.

Main Methods:

• A prospective study involving 174 patients undergoing MRCP for pancreatic evaluation.
• Quantitative analysis of signal-to-noise ratio (SNR), contrast ratio (CR), and contrast-to-noise ratio (CNR) for the distal common bile duct and pancreatic cystic lesions.
• Qualitative assessment of image quality, MPD visualization, cystic lesion conspicuity, and MPD abnormalities by two radiologists.

Main Results:

• 3D-SR demonstrated superior SNR, CR, and CNR for the distal common bile duct.
• 2D-SR and 2D-CS showed better overall image quality, motion artifact scores, MPD body/tail visualization, and margin sharpness compared to 3D-SR.
• 2D-SR and 2D-CS provided higher cystic lesion conspicuity, particularly for lesions ≤5 mm.
• All techniques accurately identified diffuse MPD dilatation and focal MPD narrowing with upstream dilatation.
• 3D-SR exhibited a higher false-positive rate for focal MPD narrowing without upstream dilatation compared to 2D techniques.

Conclusions:

• 2D-SR offers the best overall MPD visualization in a surveillance cohort.
• 2D MRCP techniques, including 2D-SR, exhibited a lower false-positive rate for focal MPD narrowing, warranting further investigation.