Abstract
Diffusion MRI (dMRI) allows for investigating the structure of the human brain. This is useful for both scientific brain research as well as medical diagnosis. Since the raw dMRI data is not directly interpretable by humans, we use mathematical models to convert the raw dMRI data into something interpretable. These models can be computed using multiple different computational methods, each having a different trade-off in accuracy, robustness and efficiency. In this thesis we studied multiple different computational models for their usability and efficiency for dMRI modeling. In the end we provide the reader with methodological recommendations for dMRI modeling and provide a high performance GPU enabled dMRI computing platform containing all recommendations.
Original language | English |
---|---|
Awarding Institution |
|
Supervisors/Advisors |
|
Award date | 25 Oct 2019 |
Place of Publication | Maastricht |
Publisher | |
Print ISBNs | 9789463805438 |
DOIs | |
Publication status | Published - 2019 |
Keywords
- Optimization
- Diffusion MRI
- microstructure modeling
- GPU computing