On 23 Sep 2004, at 02:43, Fornito, Alexander wrote:
> Thanks, that's clarified things.
> Couple more questions:
> 1 - are the .img and .hdr files saved in the reg directory (e.g.,
> example_func.hdr, highres.hdr, etc.) just the original input files
Yes, so that e.g. for higher-level analysis we where to find all
> 2 - I've noticed that for each .mat file, there is the opposite, e.g.,
> there is example_func2initial_highres.mat and
> initial_highres2example_func.mat. Why is this?
These are the inverse transforms and are used if you want to go from
the initial_highres space to example_func space (e.g. to investigate
the raw data based on highres masks with Featquery)
> Thanks again!!
> -----Original Message-----
> From: Mark Jenkinson [mailto:[log in to unmask]]
> Sent: Wednesday, September 22, 2004 5:00 PM
> To: [log in to unmask]
> Subject: Re: [FSL] Extracting registered images from FEAT
> Actually only three registrations are done - those in
> steps 1, 2 and 4. Step 3 is a concatentation of 1 & 2.
> And step 5 is a concatenation of 1 & 2 & 4.
> So to check the registrations the essential ones are
> 1, 2 and 4. To generate resampled images you can use
> flirt with the -applyxfm and -init flags to just
> apply a previously saved matrix. In this mode flirt
> does not try to find a registration solution it just
> applies the saved transform.
> As for convert_xfm and slicer - convert_xfm is used for
> concatenting matrices (which is just matrix multiplication),
> inverting matrices and converting them to medx forms.
> Slicer is used to create the 2D images that are shown
> in the feat report.
> Hope this answers your questions.
> All the best,
> On 22 Sep 2004, at 00:58, Fornito, Alexander wrote:
>> Just to clarify, are you saying that each of the first 4 steps are
>> actually separate independent registrations (i.e., none of the steps
>> depends on what happens in the preceding steps, eg., the step 3
>> registration does not use any information from the step 1
>> and then the final step is a registration of the task epi to the
>> template using a concatenation of the transformation matrices from the
>> preceding 4 registrations?
>> If so, then if I wanted to check each registration step in medx (for
>> which I would require the registered .img and .hdr, then I would just
>> run a series of standard flirt registrations?
>> Can you then explain to me what the "convert_xfm" and "slicer"
>> do in the scripting?
>> Sorry to be a pain, I'm just tryaing to get a handle on exactly what
>> Thanks again,
>> -----Original Message-----
>> From: Stephen Smith [mailto:[log in to unmask]]
>> Sent: Saturday, September 18, 2004 5:03 PM
>> To: [log in to unmask]
>> Subject: Re: [FSL] Extracting registered images from FEAT
>> Hi - it's probably simpler than you're imagining - actual
>> are only run for each incremental pair of images, namely your steps
>> 1,2,3,4 are actual registrations, and are all displayed on the report
>> page. Any other transforms (eg 5) are created by mathematical
>> concatenation of matrices and don't need any further checking if 1-4
>> For any given transform, if you don't have a saved image that you want
>> evaluate, you can take an input image, a reference image, and apply a
>> saved transform with FLIRT without re-estimating the transform, using
>> -applyxfm -init <transform>
>> On Fri, 17 Sep 2004, Alex Fornito wrote:
>>> I'm registering me task EPI via a wholehead EPI (6 DOF), the person's
>>> structural (7 DOF), and then to the template (12 DOF). The FEAT
>>> gives visual overlays for 4 registrations:
>>> 1- task EPI to wholehead EPI
>>> 2- wholehead EPI to highres structural
>>> 3- task EPI to highres structural
>>> 4- highres structural to template
>>> 5- task epi to template
>>> I understand the rationale for registering the task epi to the
>>> structural via the wholehead epi, and the then to the template via
>>> highres structural, but could you please explain the steps, and what
>>> happens (e.g., which transformation matrices get applied to what and
>>> why?). I've had a bit of trouble working it out from the scripts in
>>> the .log file.
>>> Also, I've noticed FEAT spits out the .mat files, but is it possible
>>> extract the registered images, so that I can overlay them (as per the
>>> images above) in medx and check the registration in more detail?
>>> Appreciate your help,
>> Stephen M. Smith DPhil
>> Associate Director, FMRIB and Analysis Research Coordinator
>> Oxford University Centre for Functional MRI of the Brain
>> John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
>> +44 (0) 1865 222726 (fax 222717)
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
Christian F. Beckmann
Oxford University Centre for Functional
Magnetic Resonance Imaging of the Brain,
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
Email: [log in to unmask] - http://www.fmrib.ox.ac.uk/~beckmann/
Phone: +44(0)1865 222782 Fax: +44(0)1865 222717