Hi Tobias,
Your metric files were only 30 lines long -- header but no data.
Anyway, I suspect this is a case where Caret doesn't recognize your
nifti1 .img and .hdr format. It's reading it as Analyze, which works,
except you need to then set the origin. The attached captures
overlay_bad.jpg and overlay_good.jpg show how the volume intersects the
average fiducial surface before and after using Volume: Edit Volume
Attributes: to set the origin to -90,-126,-72. (I didn't double-check
that this is precisely the right origin!) Saving the volume with the
tweaked origin results in a reasonable looking metric, as the attached
capture shows.
If you can export a flavor of NIFTI that gets along better with Caret,
then perhaps a way around this problem is to do this one; save the .hdr
as something like "make_caret_happy.hdr"; and copy this .hdr over your
existing (or copied) .hdr for mapping purposes. If you do this, please
check two things: 1) The left-right orientation is good; 2) the origin
is exactly right.
For what it's worth, I've used the .nii style NIFTI files with good
success.
Donna
On 03/26/2007 08:37 AM, Tobias Egner wrote:
> Dear Donna,
>
> thanks for your help: I've uploaded a couple of volumes
> ('Tobias_feature.hdr', etc) and the metric files (note that in the
> metric files, my volumes are referred to by different names). I'm
> actually using Windows though, not Linux.
>
> Many thanks and best wishes,
>
> Tobias.
>
> On 3/26/07, Donna Dierker <[EMAIL PROTECTED]> wrote:
>> Hi Tobias,
>>
>> I don't recognize this problem. Hmmm. How about if I give it a
try on
>> my end. Can you upload one or more of your volumes here:
>>
>> http://pulvinar.wustl.edu/cgi-bin/upload.cgi
>>
>> From the "'QEventDispatcherUNIX::unregisterTimer" error, I guess
you're
>> running Linux. We'll see if my Linux box acts the same.
>>
>> Also, would you mind also uploading one of the metric files? Or
you can
>> just verify that it is all zeros like so:
>>
>> caret_file_convert -text my.metric
>>
>> cut -f2 -d' ' my.metric |sort -un
>>
>> If you get a bunch of lines with scalars in them, then the metric
file
>> contains non-zero nodes; however, if you just get a bunch of lines
with
>> alpha characters, one line with 0.000000, and one with 73730, then
it is
>> all zeroes.
>>
>> Donna
>>
>> On 03/24/2007 03:08 PM, Tobias Egner wrote:
>> > Dear all,
>> >
>> > I am trying (unsuccessfully) to display some SPM T-images
generated in
>> > SPM5 in Caret 5.5. Here's what I am doing:
>> >
>> > The T-images are in nifti1 format (with .img and .hdr
extensions), and
>> > the original functional data were converted to nifti via the DICOM
>> > import function in SPM5. The images are normalized (and
resliced) to
>> > the SPM5 MNI T1.nii template (voxel size 2x2x2, dimensions:
91x109x91,
>> > origin: 46 64 37).
>> >
>> > In Caret 5.5, I follow the steps indicated in the
>> > Caret_Tutorial_Sep22.pdf section 5.1. (and using the Caret Tutorial
>> > Sept06 data set).
>> >
>> > 1- I open the fMRI mapping template Spec file, select all, then
save
>> > under a new name, and load scenes.
>> > 2- Attributes: Map volumes to surface, then add my SPM t-maps
and set
>> > thresholds.
>> > 3- Map to spec file with atlas: As output I select my newly created
>> > spec file, and then I select the SPM2 space, and the corresponding
>> > SPM2 left hem atlas, then repeat those steps with the SPM2 right
hem
>> > atlas. I accept the default settings on the next page.
>> > 4- The I name my metric files (left, right) in accrodance with the
>> > left/right aspects of the 'surface family'.
>> > 5- As Mapping Algorithm, I accept the 'enclosing voxel'
default, then
>> > I press Finish and the computing starts...
>> >
>> > At the end of the number crunching, I get a little warning sound
and
>> > in the command line window it reads:
>> >
>> > 'QEventDispatcherUNIX::unregisterTimer: invalid argument'
>> >
>> > Regardless, I next load the Scene No 4 (FLAT etc..), and then
open the
>> > two metric files I have created. I then also load the SPM2 fiducial
>> > coord files. I select Metric as the Primary Overlay, and select
L-to-L
>> > R-to-R matching. With great anticipation, I select one of my metric
>> > files, but no cigar: I don't see any surface overlays whatsoever.
>> >
>> > Any idea where I am going wrong? Many thanks for any advice!
>> > Best wishes,
>> >
>> > Tobias.
