Deep Learning for Mapping of Electromagnetic Tissue Properties
Member-Initiated Symposium

ORGANIZERS: Ulrich Katscher, Matthew Cronin

Wednesday, 15 May 2019

Session Number: MIS-18

Overview
This Symposium was proposed by the Electro-Magnetic Tissue Properties (SWI) study group.

"Deep Learning for Mapping of Electromagnetic Tissue Properties" gives an overview of recent developments of utilizing deep learning (DL) techniques for mapping magnetic susceptibility (Quantitative Susceptibility Mapping, QSM) and electric conductivity (Electrical Properties Tomography, EPT). Machine learning in general and deep learning in particular offer unique chances to improve image acquisition, reconstruction, and interpretation, particularly for MRI. At this annual meeting, several presentations are scheduled on machine learning "from basics to the clinic" at the Tuesday morning plenary session. In line with the corresponding study group's mission, the symposium was constructed to focus this topic on electromagnetic tissue properties mapping (EMTP). Its relevance and timeliness are underlined by several recent studies exploring this area, which will be reviewed during the symposium, thus
offering the chance for increased understanding and advancing this field.

Target Audience
Scientists, engineers, clinical researchers and radiologists interested in developing and applying DL-based susceptibility/conductivity mapping.

Educational Objectives
As a result of attending this course, participants should be able to:
- Identify the main concepts underlying deep learning, and the opportunities and challenges applying deep learning to medical imaging;
- Recognize the main concepts underlying mapping of magnetic susceptibility, electric conductivity, and permittivity, as well as drawbacks of current implementations of these techniques; 
- List opportunities and pitfalls of applying deep learning techniques to Quantitative Susceptibility Mapping (QSM) and Electric Properties Tomography (EPT), particularly with respect to the complexity of forward models involved; and
- Describe the view of radiologists on applying deep learning techniques for QSM and EPT.