CAI Seminar Series: Progress Report Seminar on DeepQSM
Event Details
- Date:
- Friday, 30 November 2018
- Time:
- 3:00 pm - 4:00 pm
- Room:
- Level 2 Seminar room
- Location:
- Building 57, Centre for Advanced Imaging, St Lucia campus
- URL:
- https://cai.centre.uq.edu.au/events
- Event category(s):
Event Contact
- Name:
- Miss Maria Moran
- Phone:
- 3365 1785
- Email:
- m.moran3@uq.edu.au
- Org. Unit:
- Science
Event Description
- Full Description:
- Progress Report Seminar on DeepQSM - Hosted by Dr Steffen Bollmann.
Quantitative susceptibility mapping (QSM) extracts magnetic susceptibility from MR signal phase and has the potential to give further insight into neurodegenerative diseases.
Background field removal is a fundamental step in the QSM processing pipeline with the goal to remove contributions of the magnetic field from outside the region of interest. The inverse problem relating the measured field perturbation to the underlying magnetic susceptibility distribution can then be solved to provide the final quantitative susceptibility map. Deep learning has been shown to be a powerful method for solving the field-to-source inversion problems of QSM without the use of regularization parameters during prediction by learning the forward field solution and efficiently solving the inverse problem. We developed a new algorithm to remove the background field from GRE MR phase images based on simulations of realistic field configurations and show that we can solve the problem in a very efficient way.
QSM algorithms based on deep convolutional neural networks have shown to produce artefact-free susceptibility maps. However, clinical scans often have a large variability, and it is unclear how a deep learning-based QSM algorithm is affected by discrepancies between the training data and clinical scans. We investigated the effects of different B0 orientations and noise levels of the tissue phase on the final quantitative susceptibility maps.
Presenters: Pia Christine Høy, Matilde Holm Kristensen, Morten Skaarup Larsen, Mathias Vassard Olsen, Mads Jozwiak Pedersen and Kristine Storm Sørensen
Event Tools
Share This Event
Print
Email
Share
