What type of study is this?

September 17, 2025

Deep learning reconstruction for improving the image quality of reduced-field-of-view diffusion-weighted imaging of the pancreas

Key Points

Deep learning-based diffusion-weighted imaging significantly reduced noise and distortion in images, enhancing clarity.
Using deep learning in reduced-field-of-view imaging increased contrast ratios between pancreatic fat and lesions, improving diagnostics.
Qualitative analysis on a 4-point scale showed marked improvements in full-size and reduced-field-of-view imaging with deep learning.
Combining reduced-field-of-view imaging and deep learning techniques led to clearer delineation of pancreatic tumors, suggesting a major advancement in imaging.

Abstract

Motivation: Deep Resolve techniques would improve DWI image quality. Goal(s): To investigate the feasibility of DR-based full-size and reduced FOV DWI (fDWI and rDWI), compared with the fDWI and rDWI. Approach: In 66 patients, fDWI, DR-fDWI, rDWI and DR-rDWI data sets were analyzed qualitatively using a 4-point scale. Contrast ratios between the pancreatic parenchyma and fat tissue or pancreatic lesion were also measured. Results: DR-based fDWI and rDWI reduced noise and distortion, sharply delineate pancreas edges, and improve overall image quality. Especially, the use of DR in rDWI increased the contrast of the pancreas and clearly delineated PDACs and NEN. Impact: Reduced FOV DWI technique decreases ghosting and susceptibility artifacts and increases spatial resolution, but decreases the signal of pancreas. Combination of reduced FOV DWI technique and DR decreased noise, improved CR pancreas-fat, and clearly delineated PDACs and NEN.

Bookmark

Deep learning reconstruction for improving the image quality of reduced-field-of-view diffusion-weighted imaging of the pancreas

Key Points

Abstract

Cite This Study