Back to the list
Congress: ECR25
Poster Number: C-13492
Type: Poster: EPOS Radiologist (scientific)
Authorblock: F. Pivetta, J. Kolck, D. Geisel; Berlin/DE
Disclosures:
Fabio Pivetta: Nothing to disclose
Johannes Kolck: Nothing to disclose
Dominik Geisel: Nothing to disclose
Keywords: Abdomen, Artificial Intelligence, Pancreas, MR, MR-Cholangiography, Comparative studies, Cysts, Neoplasia
Purpose

TThe prevalence of asymptomatic pancreatic cysts is estimated at 8%. However, due to the widespread utilization of cross-sectional imaging and continuous enhancements in image quality, the detection of incidental pancreatic cysts has surged over recent decades. Up to 44.7% of magnetic resonance cholangiopancreatography (MRCP) scans reveal pancreatic cystic lesions these, intraductal papillary mucinous neoplasms (IPMN) are particularly common. IPMNs can be categorized into three types based on imaging and / or histology: main duct IPMN (MD-IPMN), branch duct IPMN (BD-IPMN) and mixed type. All types carry the potential to undergo malignant transformation, with a pronounced risk in MD-IPMN of up to 70%. The criteria for resecting branch duct IPMN have shifted over time from early intervention to more cautious observation. For low-risk lesions, a follow-up every six months is recommended, extending to biennial examinations in the absence of changes; this underscores the need for swift yet precise assessments. Our study sought to evaluate the effectiveness of a deep learning half-Fourier single-shot turbo spin-echo (DL HASTE) sequence in comparison to a conventional HASTE for 3T-MRI of IPMN. Our hypothesis suggests that the DL HASTE sequence is not inferior in quality or quantity, despite its shorter acquisition time.

GALLERY