Works well.
Thanks Mark!
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Mark Jenkinson
Sent: Thursday, October 18, 2007 4:30 AM
To: [log in to unmask]
Subject: Re: [FSL] MNI Space in MRIcro and FSL
Dear Matt and Ged,
Sorry for not getting involved in this thread earlier but I was out of
contact.
There is a very simple solution to this.
The script I had previous sent is now outdated and the functionality has
been
incorporated directly into the FSL4 version of flirt with the -usesqform
option.
So all you need to do is take your original command:
flirt -in 4X4X4mmROI -ref MNI152_T1_1mm_brain -applyxfm -init eye.mat
-out 1X1X1mmROI
and change it to:
flirt -in 4X4X4mmROI -ref MNI152_T1_1mm_brain -applyxfm -usesqform -out
1X1X1mmROI
This should then align the (0,0,0) mm coordinates as required, for
either nifti or analyze
files.
All the best,
Mark
DRC SPM wrote:
>Hi Matt,
>
>Thanks for sending your data, the problem is that Mark's script
>assumes the data is NIfTI, and tries to grep the q-/s-form from the
>output of fslhd, but with Analyze images fslhd doesn't try to print
>this information. If you use `fslchfiletype NIFTI_PAIR your_ROI` then
>fslchfiletype will guess basic q-/s-forms (as I assume MRIcron, SPM,
>etc. also do) and then flirt_qsinit will work on this image.
>
>Sorry, I'd lost the scent a bit, and forgot that your original message
>asked about Analyze...
>
>The script could be changed to use fslhd -x, which prints an sform,
>even for Analyze data, though it would need re-formatting; or to use
>fslchfiletype and a temporary file, but I haven't bothered doing
>either of these things, mainly because it looks like I wouldn't be
>able to post the modified script to the list...
>
>I gather there is to be an FSL Extensions page on nitrc.org soon, so
>maybe the script should end up there? But since it's Mark's rather
>than mine, I don't think I should submit it there under my account.
>
>Anyway, at least we've worked out the problem, if not where to put the
solution!
>
>Ged.
>
>
>On 17/10/2007, Matt Glasser <[log in to unmask]> wrote:
>
>
>>Hi Ged,
>>
>>I was running the script on ubuntu, but shell is set to bash (not dash).
>>The error it gave was "cannot find concat matrix" which made me think
there
>>was something wrong with the script. It did that no matter what I set for
>>the s/qform settings at the end. It did output a matrix, but this matrix
>>was not correct reduced the data in size drastically, and put it in the
>>wrong place.
>>
>>I would be happy to test your version of the script if you send it to me
>>personally, rather than through the list so it doesn't get blocked.
Perhaps
>>I made some mistake when converting the script from avw to fsl (I did it
>>with a find and replace).
>>
>>Peace,
>>
>>Matt.
>>-----Original Message-----
>>From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
>>Of DRC SPM
>>Sent: Wednesday, October 17, 2007 8:50 AM
>>To: [log in to unmask]
>>Subject: Re: [FSL] MNI Space in MRIcro and FSL
>>
>>Sorry, jiscmail rejected the attachment for some reason...
>>
>>On 17/10/2007, DRC SPM <[log in to unmask]> wrote:
>>
>>
>>>Hi Matt,
>>>
>>>I wonder why it didn't work? I've just run it through avw2fsl (and
>>>also changed /bin/sh to /bin/bash in the #! line for good measure, as
>>>I seem to recall Debian/Ubuntu was unhappy with sh), and tested it on
>>>a pair of images with different image dimensions and different q- and
>>>s-forms (from co-registration in SPM5) as follows, and the result
>>>looks fine.
>>>
>>>flirt_qsinit.sh src ref src2ref.mat
>>>flirt -in src -ref ref -out src2ref -init src2ref.mat -applyxfm
>>>fslview ref src2ref
>>>
>>>I've attached my copy in case it helps, if it doesn't, hopefully Mark
>>>or Steve could take a look at your data if you upload it? (Or I could,
>>>if you put it somewhere I can see it).
>>>
>>>Best,
>>>Ged
>>>
>>>
>>>On 17/10/2007, Matt Glasser <[log in to unmask]> wrote:
>>>
>>>
>>>>Thanks Ged.
>>>>
>>>>I couldn't get the script to work, even when I changed avw to fsl, but
>>>>
>>>>
>>the
>>
>>
>>>>SPM method worked fine.
>>>>
>>>>Peace,
>>>>
>>>>Matt.
>>>>
>>>>-----Original Message-----
>>>>From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>>>
>>>>
>>Behalf
>>
>>
>>>>Of Ged Ridgway
>>>>Sent: Tuesday, October 16, 2007 10:21 AM
>>>>To: [log in to unmask]
>>>>Subject: Re: [FSL] MNI Space in MRIcro and FSL
>>>>
>>>>Hi Matt,
>>>>
>>>>Mark J posted a script that can use the NIfTI header info to generate
>>>>a FLIRT transformation:
>>>> http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0701&L=FSL&P=R20004
>>>>I think if you apply this transformation with flirt, with the standard
>>>>as the -ref, then your FSLview results should match your MRIcron ones.
>>>>
>>>>I'm guessing MRIcron resamples a binary ROI in the space of the
>>>>standard image, with either nearest neighbour interpolation or linear
>>>>followed by re-binarisation at a threshold of 0.5 -- you might need to
>>>>find out its implementation details to exactly reproduce the results.
>>>>Possibly MRIcron has an option to write out the resampled ROI, that it
>>>>presumably stores internally for display purposes?
>>>>
>>>>Hope that works,
>>>>Ged.
>>>>
>>>>P.S. Another option, if you use SPM (sorry to mention the enemy on
>>>>this list again!), is to use "coreg"->"reslice" with the standard
>>>>image defining the space and the ROI in the "other images" (or words
>>>>to that effect). The resulting ROI images prefixed with "r" should
>>>>then overlay properly in FSLview. Check out spm_reslice if you want to
>>>>avoid using the GUI.
>>>>
>>>>
>>>>Matt Glasser wrote:
>>>>
>>>>
>>>>>Hi Steve,
>>>>>
>>>>>How do you suggest I go about correcting for this difference? I could
>>>>>
>>>>>
>>>>just
>>>>
>>>>
>>>>>do it with nudge, but I would prefer an automated way that does not
>>>>>
>>>>>
>>have
>>
>>
>>>>the
>>>>
>>>>
>>>>>chance of human error.
>>>>>
>>>>>Thanks,
>>>>>
>>>>>Matt.
>>>>>
>>>>>-----Original Message-----
>>>>>From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>>>>
>>>>>
>>Behalf
>>
>>
>>>>>Of Steve Smith
>>>>>Sent: Tuesday, October 16, 2007 8:27 AM
>>>>>To: [log in to unmask]
>>>>>Subject: Re: [FSL] MNI Space in MRIcro and FSL
>>>>>
>>>>>Hi,
>>>>>
>>>>>On 16 Oct 2007, at 12:58, Matt Glasser wrote:
>>>>>
>>>>>
>>>>>
>>>>>>Hi Steve,
>>>>>>
>>>>>>Well okay, but why would the ROIs line up with FSL's standard brain
>>>>>>
>>>>>>
>>in
>>
>>
>>>>>>MRIcro, but not in FSL itself? I used the same standard space
>>>>>>image in both
>>>>>>programs
>>>>>>
>>>>>>
>>>>>Ah - I hadn't appreciated that - sorry...
>>>>>
>>>>>
>>>>>
>>>>>>and it was only misaligned in one of them, so I don't know how I
>>>>>>could use a flirt transform in this case.
>>>>>>
>>>>>>
>>>>>Right - in that case the difference presumably is because MRICRO uses
>>>>>the co-ordinate centre from the header in determining display
>>>>>centering, whereas FSLView only uses the raw voxels for working out
>>>>>display (the co-ordinate centre is only used in the reporting of the
>>>>>mm co-ordinates, not in the image display).
>>>>>
>>>>>Cheers, Steve.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>>Thanks,
>>>>>>
>>>>>>Matt.
>>>>>>
>>>>>>-----Original Message-----
>>>>>>From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>>>>>Behalf
>>>>>>Of Steve Smith
>>>>>>Sent: Tuesday, October 16, 2007 2:00 AM
>>>>>>To: [log in to unmask]
>>>>>>Subject: Re: [FSL] MNI Space in MRIcro and FSL
>>>>>>
>>>>>>HI - I presume this is because of a change in field-of-view in the
>>>>>>different standard space images. One easy way to solve that is to use
>>>>>>flirt to estimate the transform between the standard space image in
>>>>>>SPM/mricro and the FSL standard space image, and use that instead of
>>>>>>eye.mat in your call to flirt resample the ROI.
>>>>>>
>>>>>>Cheers, Steve.
>>>>>>
>>>>>>
>>>>>>On 16 Oct 2007, at 00:17, Matt Glasser wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>>Hi,
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>I have a set of ROIs from SPM in MNI space that I want to load into
>>>>>>>FSL. When I take the standard FSL brain MNI152_T1_1mm_brain and
>>>>>>>load it in MRIcro and overlay the ROI, it looks fine, like the
>>>>>>>first screenshot: http://www.ma-tea.com/~Matt/MRIcroOnMNI.png
>>>>>>>However, if I resample the 4X4X4mm ROI so it can be overlayed on
>>>>>>>this brain in FSLView, using:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>flirt -in 4X4X4mmROI -ref MNI152_T1_1mm_brain -applyxfm -init
>>>>>>>eye.mat -out 1X1X1mmROI where eye.mat is the identity matrix:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>1 0 0 0
>>>>>>>
>>>>>>>0 1 0 0
>>>>>>>
>>>>>>>0 0 1 0
>>>>>>>
>>>>>>>0 0 0 1
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>The result is incorrect, shown by the second screenshot in FSLView:
>>>>>>>http://www.ma-tea.com/~Matt/FSLOnMNI.png
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>The ROI file header is the following:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>><nifti_image
>>>>>>>
>>>>>>> nifti_type = 'ANALYZE-7.5'
>>>>>>>
>>>>>>> image_offset = '0'
>>>>>>>
>>>>>>> ndim = '4'
>>>>>>>
>>>>>>> nx = '41'
>>>>>>>
>>>>>>> ny = '48'
>>>>>>>
>>>>>>> nz = '35'
>>>>>>>
>>>>>>> nt = '1'
>>>>>>>
>>>>>>> dx = '4'
>>>>>>>
>>>>>>> dy = '4'
>>>>>>>
>>>>>>> dz = '4'
>>>>>>>
>>>>>>> dt = '1'
>>>>>>>
>>>>>>> datatype = '16'
>>>>>>>
>>>>>>> nbyper = '4'
>>>>>>>
>>>>>>> byteorder = 'LSB_FIRST'
>>>>>>>
>>>>>>> cal_min = '0'
>>>>>>>
>>>>>>> cal_max = '1'
>>>>>>>
>>>>>>> descrip = 'FSL4.0'
>>>>>>>
>>>>>>> sform_code = '2'
>>>>>>>
>>>>>>> sto_xyz_matrix = '-4 0 0 80 0 4 0 -112 0 0 4 -52 0 0 0 1'
>>>>>>>
>>>>>>> num_ext = '0'
>>>>>>>
>>>>>>>/>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>I would appreciate any assistance with this as I am unfamiliar with
>>>>>>>this type of issue if it has to do with the analyze file format or
>>>>>>>header.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>Thanks,
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>Matt.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>>
>>----------------------------------------------------------------------
>>
>>
>>>>>>--
>>>>>>---
>>>>>>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
>>>>>>
>>>>>>
>>>>>>
>>----------------------------------------------------------------------
>>
>>
>>>>>>--
>>>>>>---
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>>
>>------------------------------------------------------------------------
>>
>>
>>>>>---
>>>>>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
>>>>>
>>>>>
>>>>>
>>------------------------------------------------------------------------
>>
>>
>>>>>---
>>>>>
>>>>>
>>>>>
>>>
>>>
|
