In this research area novel methodological approaches are developed for working with extremely large databases of MR images, or images and genetics, and the complex statistics required in neuroimaging.
Extremely large datasets, of thousands of images and higher (e.g. 100,000 for the proposed UK Biobank) are hugely valuable resources for neuroimaging but specialised methods are required to work with such large amounts of data. Similarly, studies of imaging genetics need to be large and also have complicated relationships within the data.
In this research areas novel methodological approaches are developed for several goals: (i) working with large datasets; (ii) working with data involving complex relationships (e.g. imaging genetics); and (iii) more flexible, robust and sensitive statistical methods. Researchers working in this area cut across many different areas of neuroimaging and span research in efficient informatics through to fundamental statistics.
Work in this area plays a crucial part in many major ongoing projects such as the Human Connectome Project, the Developing Human Connectome Project, UK Biobank and imaging genetics studies such as ENIGMA. Methods that are developed are released as part of FSL as well as plugins and specialised pipelines for these major projects.
FSL Tools
- Randomise
- PALM
GPU implementations of:
- BEDPOST, PROBTRACKX, EDDY
Project Pipelines
Tools are implemented in the following pipelines for the automated analysis of large datasets:
- HCP
- dHCP
- Biobank UK