Project 4: Data driven model adaptions for coil sensitivities in MR systems

 

Project Partners:
University of Bremen, Germany
SagivTech, Israel

 

Project Scientists:

ESR04
José Carlos Gutiérrez Pérez

 

 

Peter Maaß
University of Bremen
FB 3 Mathematik und Informatik
Zentrum für Technomathematik
Postfach 330 440
28344 Bremen

Email: pmaass@math.uni-bremen.de
Tel.: +49-421-218-63 801

 

Tobias Kluth
University of Bremen
FB 3 Mathematik und Informatik
Zentrum für Technomathematik
Postfach 330 440
28344 Bremen

Email: tkluth@math.uni-bremen.de
Tel.: +49(421) 218 – 63817

 

Chen Sagiv
SagivTech
Hatidhar st. 16,
P.O. Box 2622,
Ra’anana, 4365104 Israel

Email: chen@sagivtech.com
Tel: +972-9-7411561

 

 

 

Project Description:

Magnetic particle imaging (MPI) is an evolving Technology aiming at non-radiative, non-invasive imaging of functional parameters such as blood flow or targeted metabolic processes. In particular, reconstruction quality is limited due to the restricted approximation Quality of PDE-based models. Data-driven approaches, based on neural networks and deep learning, would allow to incorporate expert information obtained from experimental measurements and to improve diagnostic potential of MPI technology.

Objectives of this research project are the analysis of limitations of PDE-based models (Maxwell and derived models) for coil sensitivities with a wide range of further applications. The work comprises development of concepts for data-driven operator adaptions under efficiency constraints as well as the implementation of deep learning methods for model adaptions.