Meeting Banner
Abstract #2198

DeepQSM - Solving the Quantitative Susceptibility Mapping Inverse Problem Using Deep Learning

Mads Kristensen1, Kasper Gade Bøtker Rasmussen1, Rasmus Guldhammer Blendal1, Lasse Riis Østergaard1, Maciej Plocharski1, Andrew Janke2, Christian Langkammer3, Kieran O’Brien2,4, Markus Barth2, and Steffen Bollmann2

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

Quantitative susceptibility mapping (QSM) aims to extract the magnetic susceptibility of tissue by solving an ill-posed field-to-source-inversion. Current QSM algorithms require manual parameter choices to balance between smoothing, artifacts and quantitation accuracy. Deep neural networks have been shown to perform well on ill-posed problems and can find optimal parameter sets for a given problem based on real-world training data. We have developed a proof-of-concept fully convolutional deep network capable of solving QSM’s ill-posed field-to-source inversion that preserves fine spatial structures and delivers accurate quantitation results.

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

Join Here