Hi,
> Thanks for your quick response. Yes, your interpretation of my second
> question is right. I have 6 runs for each subject and am combining
> them at
> the second level to see activations for this subject. When you say
> check
> higher-level registration report, I am not sure what you are
> referring to. I
> have a folder called inputreg which has the overlap images and
> those look
> good to me.
That's right, and there's a link to a webpage showing those results
from the main higher-level output webpage (*.gfeat/report.html)
> Also, I just specify stats+post-stats option for higher-level
> and the registration is turned off here. In my fsf file, I have "Set
> fmri(reg_yn) 0". Should this not be the case and should this be 1
> instead?
No - there is no further registration to do once the lowest-level
FEAT runs have been done - from that point on everything will be in
standard space (however that was defined at the lowest level).
> My assumption was that since all images are registered in the lower
> level to
> the standard space, there is no need to do run registration at the
> higher
> levels except apply the estimated transformation values to the
> statistical
> images in normal space. Is this correct?
Correct.
> Also, I am not sure what do you mean by "look at the registration
> of the two
> example_funcs to each other". Are you referring to just overlaying
> them on
> each other to see if they match? I have tried to do this visually
> using
> fslview and there does seem to be some difference between the two
> images
> (esp when comparing run1 and run6). But I am not able to quantify
> this in
> some way or attribute it to poor registration. The summary of
> registration
> output from lower level look similar to me (see my related question
> below).
> And if it is indeed a registration issue, am not sure what I can do to
> improve it.
I would simply judge whether the standard space versions of the
example_func images (as seen in the summary reports described above)
look good and similar to each other.
> Finally, how should one interpret the registration outputs from FSL
> systematically? Any help on this would be useful. I am not sure
> what the red
> areas represent in these outputs.
The red lines show the intensity edges from one image superimposed on
the other image.
> If I have already run a full-model at the lower level and want to
> change
> some aspects of my design and run a new model on the same data, I
> understand
> that I can avoid having to do preprocessing again by using the
> filtered_func
> data saved in the previous model. Is there a similar way in which I
> can tell
> FSL to read the registration values from the previous analysis
> rather than
> have to perform flirt again?
No, sorry. However there is a danger is re-using parts of previous
analyses, which is that you end up with a possibly messy set of
directories and files, and end up getting confused. I generally
recommend always running FEATs from scratch, even though that takes
longer to run, as things end up neater, which in the long run is very
important.
Cheers.
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
------------------------------------------------------------------------
---
|