Clusterwise Independent Component Analysis for multi-subject (resting state) fMRI data
Our brain is a network of functionally connected brain regions. Functional network integrity may disrupt in the cascade of events leading to dementia before anatomical or cognitive changes become apparent. Increasing sensitivity to identify network alterations therefore may enhance the early detection of dementia.
To this end we propose Clusterwise Independent Component Analysis (CICA), a technique for multi-subject fMRI data that clusters subjects based on within-cluster similarities and between-cluster differences in functional networks. As such, networks sensitive to early changes may be discovered in a data-driven way. Additionally, we will also study how anatomical MRI information can be incorporated into the technique.
To facilitate applied researchers to use our novel CICA method, a user-friendly R package will be developed. And since applied researchers may not be very familiar with R, we will also create a web application (by making use of R shiny apps) to make the CICA software easy accessible.
Dr Tom F. Wilderjans & Prof. Serge A.R.B. Rombouts
NWO Research Talent Grant
1 September 2016 1 September 2021