 |
 |
FAST On 9 Nov 2012, at 13:30, Rongxiang Tang wrote: > Hi Jeanette, > > Thanks for the long response. > > I'm trying to figure out how to create the mask for WM and CSF....and waiting to see if others know how to do it. > > Thanks again, > Catherine > > From: Jeanette Mumford <[log in to unmask]> > To: [log in to unmask] > Sent: Thursday, November 8, 2012 7:06 PM > Subject: Re: [FSL] how to calculate and regress out WM and CSF confounds in fsl > > Hi Catherine, > > > Using the approach of your choice, you would first create masks for WM and CSF. Somebody else can comment on what may be the best way to do this. > > Next, put each mask in a separate file (if not already) such that it is a binary mask, taking a value of 1 within the CSF and 0 everywhere else (repeat for WM). I'm not sure if fast does this by default or if it creates a single file where the voxel intensity is an integer, say 1, 2 or 3, and that indicates which type of matter that voxel is (you can find that on the FSL web page or by looking at the output image). You can use fslmaths img -thr 2 -uthr 2 -bin mask, to create a mask of the voxels that have 2. Again, I don't know what fast outputs because I haven't looked at it in a long time. > > Once you have a binary mask, you need to transform it to the BOLD data space. You would use flirt to do this. You would need the mat file that goes from the subject's T1 to the BOLD data space, but you start by finding the bold to T1. > flirt -in bold.nii.gz -ref T1_image -omat bold2T1.mat -dof 6 > Then you need the inverse of the bold2T1.mat file, which you'd get using the InvertXFM_gui. It is a simple GUI. Just input the bold2T1.mat file and call the inverse T12bold.mat > > Then you need to move your ROI mask into BOLD space using R12bold.mat > > flirt -in CSF_mask -ref bold -applyxfm -init T12bold.mat -out CSF_inboldspace -interp nearestneighbour > > Now you have your mask in the same space as your bold data (check in in fslview to make sure it looks good. It should be binary). Now you can extract your time series > > fslmeants -i bold -m CSF_inboldspace -o CSF_timeseries.txt > > Now you're all set to stick that into your feat design matrix to adjust for it in your analysis. > > Hope that helps, > Jeanette > > On Thu, Nov 8, 2012 at 6:08 PM, Rongxiang Tang <[log in to unmask]> wrote: > Dear All, > > I was wondering how to calculate and regress out the WM and CSF confounds in fsl... > > I assume that one would first run FAST on all subjects' brain extracted T1 image...and get the output wm images, etc....but what's next? > > How do you get the signals...(may into a txt file?), and regress them out? do I also need to set a threshold for the wm or csf image? > > It would be much appreciated if someone could explain the steps in detail.... > > Thanks in advance, > Catherine > > > --------------------------------------------------------------------------- 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 ---------------------------------------------------------------------------