Thanks John, that worked perfectly. Just one further question - you
suggest that after deskulling I coregister ...
> the functional data to the output of ImCalc.
> This would be done by having the ImCalc result as "reference", one of
> the fMRI scans as the "source", and all the others as "other".
... and use the resulting ***_seg_sn.mat for subsequent MNI normalisation.
Isn't it usually more straightforward to use the structural image as
source (in this case the ImCalc result as dependency), and a
representative functional image as reference (here the mean realigned
one)? Then I can put the original unskulled structural image as "other".
best,
Brian
> sJohn Ashburner wrote:
> Sometimes there is a bit too much extra information in the anatomical
> scans, which can confuse the (normalised) mutual information
> coregistration. Here is a recycled response (
> https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0906&L=SPM&P=R49092
> ), which may help.
>
> Probably the easiest fix for this would be to use the Segment button
> (maybe the New Segment of SPM8) to generate native space tissue class
> images of GM, WM and CSF, as well as a bias corrected version of your
> anatomical scan. The c1, c2 and c3 images could be used to skull strip
> the bias corrected anatomical scan, which should give you an image that
> coregisters rather better.
>
> The skull stripping bit can be done using ImCalc, where you would select
> the m***.img, c1***.img, c2***.img and c3***.img. The expression to
> evaluate would then be i1.*((i2+i3+i4)>0.5) .
>
> If you plan to use the ***_seg_sn.mat file for spatially normalising,
> then you should coregister the functional data to the output of ImCalc.
> This would be done by having the ImCalc result as "reference", one of
> the fMRI scans as the "source", and all the others as "other".
>
> Best regards,
> -John
>
>
> On 8 November 2010 09:48, Brian Murphy <[log in to unmask]> wrote:
>
>> Hi,
>>
>> I have one participant session that will not coregister for me correctly. I
>> am using the same preprocessing procedure as worked well on two previous
>> participants, so I don't think there is an error in the script.
>>
>> The result I am getting is that the structural image is rolled a little to
>> the left (~0.05 rad), pitched forward (~.1-.2 rad) and displaced downwards.
>> Functional realignment is working fine, and I am using the resulting mean
>> image as the reference.
>>
>> The things I have tried so far are to make the tolerances smaller, to move
>> the origin in the structural image on to the AC (using the Display
>> function), to approximately align the functional and structural images by
>> hand (with Check Reg), and combinations of these. But always when I re-run
>> coregistration I get the same result.
>>
>> I guess it might have something to do with my functional scanning parameters
>> - I'm scanning broad coverage of cortex (just missing a bit of motor and
>> temporal) on a short TR (1s) and coarse resolution (3x3x5mm). Is it
>> generally difficult to get such low resolution, low contrast images to
>> coregister correctly?
>>
>> (BTW on this third participant, I'm using dummy scans. On the previous two
>> that coregistered well, I did record the initial high-contrast scans. Since
>> I am using the mean of ~400 functional images as reference, I guess those
>> few initial scans shouldn't make a difference, but I thought I should
>> mention it just in case).
>>
>> Of course I can simply coregister this participant by hand, but it would be
>> more satisfying to understand why it is going awry. Any suggestions?
>>
>> thanks,
>>
>> Brian
>>
>>
>> --
>> Brian Murphy
>> Post-Doctoral Researcher
>> Language, Interaction and Computation Lab
>> Centre for Mind/Brain Sciences
>> University of Trento
>> http://clic.cimec.unitn.it/brian/
>>
>>
--
Brian Murphy
Post-Doctoral Researcher
Language, Interaction and Computation Lab
Centre for Mind/Brain Sciences
University of Trento
http://clic.cimec.unitn.it/brian/
|