Meeting Banner
Abstract #0016

A disentangled representation trained for joint reconstruction and segmentation of radially undersampled cardiac MRI

Tobias Wech1, Julius Frederik Heidenreich1, Thorsten Alexander Bley1, and Bettina Baeßler1
1Department of Diagnostic and Interventional Radiology, University Hospital Würzburg, Würzburg, Germany


The network we propose in this work (xSDNet) jointly reconstructs and segments cardiac functional MR images which were sampled below the Nyquist rate. The model is based on disentangled representation learning and factorizes images into spatial factors and a modality vector. The achieved image quality and the fidelity of the delivered segmentation masks promise a considerable acceleration of both acquisition and data processing.

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

Join Here