I am using FSL to perform resting state analysis, just past pre-processing, I realized that identifying noise components manually is not only time comsuming, but may not be 100% accurate. Are there better automated ways (or is manual check still the securest?) and which toolboxes work well? I read about ART and FIX... What is your experience? What would you recommend?
Hi Valeriya,
I have worked with the ART toolbox but in task-related functional neuroimaging data, not in resting state. In the case of motor studies, it is usually employed as ART offers an automated way to check out and discard the artifacts and also to display the statistics of the data signal (i.e., evolution across time). I guess it would be also useful in the case of resting state data, in the case that you want to discard artifact volumes previously to the statistical analyses.
Hi! Thanks! Have you by any chance worked with FSL processed data in ART?
It says that it's supposed to be able to read 4DNifti, but I can't find the add on that does it... instead I tried to convert back into 3D in SPM8 format, which worked, but the output doesn't look quite right... (see attachment) Did I miss something in the settings or does a 4D to 3D conversion mess with the data?
I would recommend using GroupICA tool and it's recently released back-reconstruction method called 'Group-Information-Guided ICA'
Removing noise components from fMRI datasets is very challenging. Research shows that no method is perfect. This is particularly a sensitive case for resting-state studies, where we do not really know what these resting-state networks actually mean. Physiological measurements, such as a breathing belt and/or PulseOx, provide some means to removing artifacts from the data. Single-subject ICA decomposition is hit or miss. Sometimes it works well, and sometimes not so much. I wish I had more encouraging news :(
I'd definately vote for FSL FIX. Maybe you could get a trained data-set already for scanner spesific cleaning. If not then do the training yourself and then you know what will be removed.
- Badges
- Science topic
- Similar topics
- Psychology
- Cognitive Science