Abstract #2593

Deep Learning for Liver Segmentation and Quantification of Obese Patients

Philipp Madörin¹, Xeni Deligianni^1,2, Francesco Santini^1,2, Simon Andermatt², Philippe Claude Cattin², Anne Christin Meyer-Gerspach^3,4, Bettina Karin Wölnerhanssen^3,4, Oliver Bieri^1,2, and Orso Pusterla^1,2,5

¹Department of Radiology, Division of Radiological Physics, University Hospital Basel, Basel, Switzerland, ²Department of Biomedical Engineering, University of Basel, Basel, Switzerland, ³St. Clara Research Ltd, St. Clara Hospital, Basel, Switzerland, ⁴University of Basel, Basel, Switzerland, ⁵Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland

Obesity is one of the greatest health risks and strongly related to fatty liver disease. Magnetic resonance imaging enables non-invasive measurement of fat-water distribution in tissue. To provide an automated evaluation of the liver volume and fat percentage, we trained a Multi-Dimensional Gated Recurrent Units network to segment multi-contrast data. The neural network was trained with a limited number of data comprising 52, 20, 10 datasets and was evaluated for liver volume and fat percentage quantification.

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

Join Here