Abstract #4931

SHARQnet - Sophisticated Harmonic Artifact Reduction in Quantitative Susceptibility Mapping using a Deep Convolutional Neural Network

Mathias Vassard Olsen¹, Morten Skaarup Larsen¹, Matilde Holm Kristensen¹, Mads Jozwiak Pedersen¹, Lasse Riis Østergaard¹, Kieran O'Brien^2,3, Christian Langkammer⁴, Markus Barth², and Steffen Bollmann²

¹Department of Health Science and Technology, Aalborg University, Aalborg East, Denmark, ²Centre for Advanced Imaging, The University of Queensland, Brisbane, Australia, ³Siemens Healthcare Pty Ltd, Brisbane, Australia, ⁴Department of Neurology, Medical University of Graz, Graz, Austria

We propose a fully convolutional neural network for background field removal in MR phase images for Quantitative Susceptibility Mapping. Our proposed method, SHARQnet, learns to solve the background field problem from theoretical simulations of background field distributions, and the results are compared to current state-of-the-art methods like SHARP, V-SHARP, and RESHARP.

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

Join Here