Hi Paul,

              If you have run Melodic ICA as part of a full FEAT analysis then no actual denoising will have occurred - to do this you need to run in Preprocessing only mode to generate the ICA report and then use fsl_regfilt to denoise ( see the FEAT user guide for details ). The additional confound EVs file will then be applied when Statistics only mode is run on the denoised data.

Kind Regards

Matthew

Ok. Thank you.

It looks as though, according the the report.html logs that this is what FEAT does is you select both ICA exploration and additional confound EVs (i.e. the .txt output from fsl_motion_outliers).

However, I'm still not clear about one part of my question.

After doing the exploratory ICA followed by confound regressor cleanup would it then also be a good idea to further remove components (say, with FIX), or is that too much?

I'm sorry to pester with this questioning, but I really want to make sure that I'm not plagued with motion confounds since these are data that include severe Alzheimer's patients.

Cheers

On Thu, Jul 24, 2014 at 5:24 AM, Stephen Smith <[log in to unmask]> wrote:

HiOn 23 Jul 2014, at 13:28, Paul Beach <[log in to unmask]> wrote:

Stephen,

Thank you for your advice.

I figured that I would try and regress out the confounders and perform individual ICA exploration using FEAT. This step would be performed on data that have been pre-processed through WM/CSF regression (and obviously the motion correction performed in fsl_motion_outliers). 

that's OK as an exploratory approach - however, normally if you are going to apply both ICA cleanup and confound-regressor cleanup, we would recommend running the ICA first.

Would you also suggest using FIX after doing this, or is that overkill?

same answer - ICA+FIX is logically the same as ICA cleanup with manual artefact component identification

Also, as for training the FIX classifier: since I have AD and healthy control subjects, should I use equal numbers of both groups in the 10 or so training subjects?

Yes I would think so.

Cheers.

Thanks again

On Wed, Jul 23, 2014 at 4:43 AM, Stephen Smith <[log in to unmask]> wrote:

Hi

On 22 Jul 2014, at 17:03, Paul Beach <[log in to unmask]> wrote:

FSL experts and users,

I'm working toward an ICA analysis on healthy controls and AD patients for the first time and thus I am very interested in making sure I appropriately deal with motion confounds.

I've been looking through past message board correspondence on this topic and I can't seem to find anything too conclusive wrt just when/where to include the fsl_motion_outliers output as a confound EV (specifically using the Power (2012) methodology).

It seems like the most recent discussion about this topic suggested doing so at the single subject level, PRIOR to doing any group ICA/dual-regression steps (i.e. doing a single-subject melodic run and including the output in the post-stats model/contrast). Is this correct?

Nearly - though I would recommend that if your main analysis is group-ICA followed by dual regression, then you should do within-subject cleanup - either using FIX, or regressing out confounds like motion-outliers, etc.

Some of the messages also discuss interest in developing a melodic-based classifier to automatically pick out/remove motion-based components. This is yet to be incorporated in FSL, correct?

This is FIX, which is released as an FSL "plugin" now.

Cheers.

What are other folks doing to address this issue?

Thanks

--

Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program

 - MSU Cognitive and Geriatric Neurology Team (CoGeNT)

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve

---------------------------------------------------------------------------

Stop the cultural destruction of Tibet

--

Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program

 - MSU Cognitive and Geriatric Neurology Team (CoGeNT)

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve

---------------------------------------------------------------------------

Stop the cultural destruction of Tibet

--

Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program

 - MSU Cognitive and Geriatric Neurology Team (CoGeNT)