interactive dimensionality reduction of neuroimaging results
Interact live with this app on shinyapps.io \
(free account, so please allow a few seconds to load).
This app allows to explore which combination of regions from an atlas yields the maximum separation between tasks/conditions/groups. The latter is estimated using different types of dimensionality reduction techniques.
The low-dimensional embedding can then be inspected to see the profiles of the single maps which were grouped together.
While it was tested here with (synthetic) fMRI data, it can be applied to any kind of modality which produces a result in a template space.
In a further development, several atlases will be available, as well as tuning parameters for different dimensionality reduction methods.