Dear Mohit,

When running ICA at single subject level, matching similar components across subjects is a big issue because, as you pointed out, not all subjects will have the same type of components.

A solution to this is to run melodic on all your subjects (multi-session ICA)

see https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/MELODIC

to derive a common set of components

And then derive the corresponding single-subject ICA and perform statistical tests using dual regression:

https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/DualRegression

Single subject ICA is mostly used for data denoising (i.e. removing noise components from the data), rather than to select components of interest.

For more info and practical examples see

http://fsl.fmrib.ox.ac.uk/fslcourse/

especially “Model-Free FMRI Analysis”

Hope it helps,

Ludovica

On 19 Apr 2017, at 4:04 pm, Dr. Mohit Saxena <[log in to unmask]> wrote:

Hello FSLers

I am pretty new to FSLs ICA
i have a question regarding selection of components:
I ran melodic with the default 30 components output and the melodic file has 30 volumes referred to as components (0-29), and then flirted it to hte patient's T1

i have two questions:
1. if a subject has a single volume that includes all the relevant brain areas we are interested in, should we consider just this volume. If yes, what about other volumes that have parts of the areas we are interested in
2. on the other hand, what ff we have a patient that has all relevant areas in one volume and another patient that has a number of volumes that have parts of the areas we are interested in
3. Is there any weighting with respect to the number of components selected per individual?

Your response is highly appreciated

Thanks & Regards

Dr. Mohit Saxena
(Ph.D Neurology)

Analysis Postdoctoral Research Assistant
Oxford Centre for Functional MRI of the Brain (FMRIB)
Nuffield Department of Clinical Neurosciences, University of Oxford
John Radcliffe Hospital
Oxford, OX3 9DU, UK
email: [log in to unmask]