Hi,
I guess it is just the high resolution of 0.9 x 0.9 x 0.9mm that is
creating too many voxels for invwarp to currently handle nicely.
In our lab it takes about 5 hours to do one with
193 x 176 x 206 at 0.93 x 1.2 x 0.93mm
so I'm not quite sure why it is so much slower at your end.
Matt's suggestion of downsampling this image to act as
your reference in the invwarp call will work though, and
there isn't a lot of loss of accuracy in going to 1.5mm resolution,
but potentially a lot of speedup.
Can you also send me your images to the upload site?
I'd like to make sure that our new, faster version can deal
with these before we release it.
All the best,
Mark
On 4 Jul 2009, at 11:38, Markus Gschwind wrote:
> Dear Mark and Matt,
> Thank you for your explanations!
> I indeed have used a cropped and betted T1, however it did neither
> match the MNI_template, nor was it very small.
> dim1 190
> dim2 210
> dim3 160
> pixdim1 0.9000000358
> pixdim2 0.8984375000
> pixdim3 0.8984375000
>
> I do not know if the invwarp call was good or not, as I killed it
> after 52 hours.
> Thank you again,
> Markus
>
>
>
> 2009/7/3 Mark Jenkinson <[log in to unmask]>
> Hi,
>
> I think there is some confusion here.
> The 5x5x5 image is actually the coefficient file containing warp field
> parameters, and is not actually an image. It is just a representation
> of the warp field that we cram into a nifti image.
>
> What is crucial for the speed of registration - especially for the
> inverse warp - is the size and resolution of the original image.
> Not the reference image, but the original one. It doesn't matter
> for inverse warp whether you used the 2mm or 1mm template
> as the reference - what counts is the size of the image that you
> are registering *to* the template. Hence we recommend
> cropping the original images to get rid of extra space around the
> head before you run anything, as this will make things quicker.
> We hope to release a much faster version soon, but in the
> meantime cropping your images will have a huge effect. Just
> remember to do it before running the initial fnirt, not after.
>
> All the best,
> Mark
>
>
>
>
>
> On 3 Jul 2009, at 16:26, Markus Gschwind wrote:
>
> Sorry, Matt, I got your question wrong.
> I used the MNI_T1_2mm_brain.nii.gz template to create the warp. It
> was the resulting ouput that had 5x5x5 resolution ( I do not why!)
>
> I just had tried the FA_to_FMRIB_FA_1mm way (TBSS algorithm), and
> then invert this warp.
> However, as the patients has a huge right frontal lesion, it did not
> work well even with the use of an additional --inmask (sparing out
> the lesion).
>
> Thanks for any help so far!
> Markus
>
>
>
>
>
> 2009/7/3 Matt Glasser <[log in to unmask]>
> I don’t think there is a 5x5x5mm MNI template, unless you made one.
> I was expecting one of two answers 1mm isotropic or 2mm isotropic.
> Which is it?
>
>
> Peace,
>
>
> Matt.
>
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf Of Markus Gschwind
> Sent: Friday, July 03, 2009 2:44 AM
>
>
> To: [log in to unmask]
> Subject: Re: [FSL] convertwarp snytax
>
>
> Hi Matt,
>
> The resolution is 5x5x5mm and it is still running (50 hours no).
>
> Still not understanding,
> Markus
>
>
>
> 2009/7/2 Matt Glasser <[log in to unmask]>
>
> What resolution MNI image did you use for the original warp?
>
>
> Peace,
>
>
> Matt.
>
>
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf Of Markus Gschwind
> Sent: Thursday, July 02, 2009 10:18 AM
> To: [log in to unmask]
> Subject: Re: [FSL] convertwarp snytax
>
>
> Dear Jesper,
>
> Just two more things to add:
>
> 1. The flirt/ fnirt call I was using is as follows:
>
> myT1=${T1Dir}/ostruct_brain.nii.gz
> T1_MNI=/usr/local/fsl/etc/standard/MNI152_T1_2mm_brain.nii.gz
> myconfig=/usr/local/fsl/etc/flirtsch/T1_2_MNI152_2mm.cnf
>
> # Start with finding an affine transform
> flirt -ref ${T1_MNI} -in ${myT1} -omat ${T1Dir}/T1_2MNI.mat -out $
> {T1Dir}/T1FlirtMNI.nii.gz
> # And non-linear
> fnirt --in=${myT1} --config=${myconfig} --aff=${T1Dir}/T1_2MNI.mat --
> ref=${T1_MNI} --cout=${T1Dir}/T1_2MNI_warpcoef --iout=${T1Dir}/
> T1_2MNI.nii.gz
>
> 2. The invwarp of T1_2MNI is running for nearly 20 hours now (3GHz
> and 16GB RAM). That seems pretty long to me.
>
> Thanks for your help!
> Markus
>
> 2009/7/2 Markus Gschwind <[log in to unmask]>
>
> Dear Jesper!
>
> Thank you for your comment.
>
> Yes, I did not see that the warpfields were not goodlooking.
>
> However I found that they have very different dimensions and
> orientation (see fslhd on them below) even though they were created
> by similar fnirt calls...
>
> Would this give an explanation?
>
> Thank you again!
> With best wishes, Markus
>
> filename T1_2_FA1_warpcoef.nii.gz
> =============================
> sizeof_hdr 348
> data_type FLOAT32
> dim0 4
> dim1 128
> dim2 128
> dim3 64
> dim4 3
> dim5 1
> dim6 1
> dim7 1
> vox_units mm
> time_units s
> datatype 16
> nbyper 4
> bitpix 32
> pixdim0 0.0000000000
> pixdim1 1.7968750000
> pixdim2 1.7968750000
> pixdim3 2.0000000000
> pixdim4 1.0000000000
> pixdim5 0.0000000000
> pixdim6 0.0000000000
> pixdim7 0.0000000000
> vox_offset 352
> cal_max 0.0000
> cal_min 0.0000
> scl_slope 0.000000
> scl_inter 0.000000
> phase_dim 0
> freq_dim 0
> slice_dim 0
> slice_name Unknown
> slice_code 0
> slice_start 0
> slice_end 0
> slice_duration 0.000000
> time_offset 0.000000
> intent Unknown
> intent_code 0
> intent_name
> intent_p1 0.000000
> intent_p2 0.000000
> intent_p3 0.000000
> qform_name Scanner Anat
> qform_code 1
> qto_xyz:1 -1.796875 0.000000 -0.000000 117.711868
> qto_xyz:2 0.000000 1.796875 -0.000000 -102.796608
> qto_xyz:3 0.000000 0.000000 2.000000 -51.474575
> qto_xyz:4 0.000000 0.000000 0.000000 1.000000
> qform_xorient Right-to-Left
> qform_yorient Posterior-to-Anterior
> qform_zorient Inferior-to-Superior
> sform_name Scanner Anat
> sform_code 1
> sto_xyz:1 -1.796875 0.000000 -0.000000 117.711868
> sto_xyz:2 0.000000 1.796875 -0.000000 -102.796608
> sto_xyz:3 0.000000 0.000000 2.000000 -51.474575
> sto_xyz:4 0.000000 0.000000 0.000000 1.000000
> sform_xorient Right-to-Left
> sform_yorient Posterior-to-Anterior
> sform_zorient Inferior-to-Superior
> file_type NIFTI-1+
> file_code 1
> descrip FSL4.0
> aux_file
>
>
> filename T1_2MNI_warpcoef.nii.gz
> ============================
> sizeof_hdr 348
> data_type FLOAT32
> dim0 4
> dim1 21
> dim2 24
> dim3 21
> dim4 3
> dim5 1
> dim6 1
> dim7 1
> vox_units mm
> time_units s
> datatype 16
> nbyper 4
> bitpix 32
> pixdim0 0.0000000000
> pixdim1 5.0000000000
> pixdim2 5.0000000000
> pixdim3 5.0000000000
> pixdim4 1.0000000000
> pixdim5 0.0000000000
> pixdim6 0.0000000000
> pixdim7 0.0000000000
> vox_offset 352
> cal_max 0.0000
> cal_min 0.0000
> scl_slope 0.000000
> scl_inter 0.000000
> phase_dim 0
> freq_dim 0
> slice_dim 0
> slice_name Unknown
> slice_code 0
> slice_start 0
> slice_end 0
> slice_duration 0.000000
> time_offset 0.000000
> intent Unknown
> intent_code 2007
> intent_name
> intent_p1 2.000000
> intent_p2 2.000000
> intent_p3 2.000000
> qform_name Scanner Anat
> qform_code 1
> qto_xyz:1 5.000000 0.000000 0.000000 91.000000
> qto_xyz:2 0.000000 5.000000 0.000000 109.000000
> qto_xyz:3 0.000000 0.000000 5.000000 91.000000
> qto_xyz:4 0.000000 0.000000 0.000000 1.000000
> qform_xorient Left-to-Right
> qform_yorient Posterior-to-Anterior
> qform_zorient Inferior-to-Superior
> sform_name Scanner Anat
> sform_code 1
> sto_xyz:1 1.113220 -0.000225 -0.000151 -6.675520
> sto_xyz:2 -0.033047 1.067810 0.293899 -29.982800
> sto_xyz:3 0.021964 -0.324781 1.023180 -41.905102
> sto_xyz:4 0.000000 0.000000 0.000000 1.000000
> sform_xorient Left-to-Right
> sform_yorient Posterior-to-Anterior
> sform_zorient Inferior-to-Superior
> file_type NIFTI-1+
> file_code 1
> descrip FSL4.0
> aux_file
>
>
> 2009/7/1 Jesper Andersson <[log in to unmask]>
>
>
> Dear Markus,
>
>
> I am having trouble with making work a convertwarp command.
>
> I have warped dti_FA to T1 and T1 to MNI_2mm.
>
> I then inverted each warp_coef and would like to combine both into
> a MNI_2FA_warpcoef in order to apply this one to pieces out of an
> MNI_Atlas.
>
> I used the following command:
>
> convertwarp --ref=dtifit/dti_FA.nii.gz --warp1=Warps/
> T1_2_FA_warpcoef.nii.gz --warp2=T1/MNI_2T1_warpcoef.nii.gz --
> out=Warps/MNI_2FA_warpcoef.nii.gz
>
>
> your command looks OK. What does the individual inverted warpfields
> look like? Are they both OK?
>
> Jesper
>
>
>
>
> --
> Dr. med. Markus Gschwind, M.D.
> Laboratory for Neurology and Imaging of Cognition
> Dept of Neurosciences
> University Medical Center (CMU)
> 1 Michel-Servet - 1211 GENEVA - CH
>
> Tel 0041 (0) 22 379 5324
> Fax 0041 (0) 22 379 5402
> email: [log in to unmask]
> http://labnic.unige.ch
>
>
>
>
> --
> Dr. med. Markus Gschwind, M.D.
> Laboratory for Neurology and Imaging of Cognition
> Dept of Neurosciences
> University Medical Center (CMU)
> 1 Michel-Servet - 1211 GENEVA - CH
>
> Tel 0041 (0) 22 379 5324
> Fax 0041 (0) 22 379 5402
> email: [log in to unmask]
> http://labnic.unige.ch
>
>
>
>
> --
> Dr. med. Markus Gschwind, M.D.
> Laboratory for Neurology and Imaging of Cognition
> Dept of Neurosciences
> University Medical Center (CMU)
> 1 Michel-Servet - 1211 GENEVA - CH
>
> Tel 0041 (0) 22 379 5324
> Fax 0041 (0) 22 379 5402
> email: [log in to unmask]
> http://labnic.unige.ch
>
>
>
>
> --
> Dr. med. Markus Gschwind, M.D.
> Laboratory for Neurology and Imaging of Cognition
> Dept of Neurosciences
> University Medical Center (CMU)
> 1 Michel-Servet - 1211 GENEVA - CH
>
> Tel 0041 (0) 22 379 5324
> Fax 0041 (0) 22 379 5402
> email: [log in to unmask]
> http://labnic.unige.ch
>
>
>
> --
> Dr. med. Markus Gschwind, M.D.
> Laboratory for Neurology and Imaging of Cognition
> Dept of Neurosciences
> University Medical Center (CMU)
> 1 Michel-Servet - 1211 GENEVA - CH
>
> Tel 0041 (0) 22 379 5324
> Fax 0041 (0) 22 379 5402
> email: [log in to unmask]
> http://labnic.unige.ch
|