Abstract #1730

Automated Assessment of Liver Parenchymal Enhancement on Hepatobiliary Phase MR Images Using a Convolutional Neural Network

Guilherme Moura Cunha¹, Kyle A Hasenstab¹, Kang Wang¹, Timo Delgado¹, Atsushi Higaki¹, Ryan L Brunsing², Alex Schlein¹, Armin Schwartzman³, Albert Hsiao¹, and Claude B Sirlin¹

¹Radiology, University of California San Diego, La Jolla, CA, United States, ²Radiology, Stanford University, Palo Alto, CA, United States, ³Department of Family Medicine and Public Health, University of California San Diego, La Jolla, CA, United States

Adequate hepatocellular enhancement (HCE) in Gd-EOB-DTPA-enhanced MRI studies can often deviate from the standard delay of 20 minutes. In this study, we proposed a fully-automated CNN-based approach for real-time assessment of HCE adequacy and retrospectively evaluated performance using 1201 T1w HBP 3D image sets from 406 unique patients. Our proposed model classified images with inadequate uptake with an AUC of 97%. With further validation, this approach could be used to identify the earliest time point HCE adequacy is achieved, potentially shortening scanning time by tailoring the exam length to the individual liver’s ability to uptake contrast.

This abstract and the presentation materials are available to members only; a login is required.

Join Here