Hi,
I assume MTR is Magnetisation Transfer Ratio - but this doesn't
explain why 12 dof works while 6 dof fails. I assume they are of
the same subject - right?
Is there anything unusual with the intensities around the edge of
the brain? And does the 12 dof result do a lot of scaling or skewing?
Can you run avscale on the transformation matrix from FLIRT and
send us the output?
All the best,
Mark
On 20 May 2009, at 22:12, Matt Glasser wrote:
> What is MTR, I don't think I know that acronym.
>
> Peace,
>
> Matt.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf
> Of Siew-Min Gan
> Sent: Wednesday, May 20, 2009 6:43 AM
> To: [log in to unmask]
> Subject: Re: [FSL] choice of registration and MNI template
>
>> Understood. Thanks for your explaination.
>
> I noticed that with registerering MTR deskulled brains to T1, I can
> only
> do so with flirt with 12dof affine and not 6 or 9 dof. May I ask why
> does
> 6DOF works for undistorted fa but not for MTR images?
>
> Thanks
>
> Siewmin
>
> I'm sorry, I think I misunderstood you somewhere along the way and
> that is
>> the source of our confusion. You are correct that the 1mm and 2mm
>> templates
>> have different matrix sizes (this is what I didn't realize you were
>> referring to). I thought you had some other version of the MNI
>> template
>> that had more empty space around it (a bigger FOV). The FOV
>> actually is
>> exactly the same, it is just the matrix that is larger in the 1mm
>> (as it
>> divides the same FOV into smaller voxels). There will be no
>> translational
>> difference in the images (I just verified this by resampling the 2mm
>> template to the 1mm template using an identity matrix). If your
>> images
>> had
>> different FOVs, rather than different matrix sizes, you cannot
>> necessarily
>> just resample the images with an identity matrix.
>>
>> If you look carefully in that config file, you will see that the last
>> subsampling step is 2 rather than 1 (and hence the highest resolution
>> reference image used is 2mm). Since you want your transform to be
>> from
>> the
>> 1mm MNI space to the 1mm T1 space (and then the FA), you could just
>> edit
>> the
>> config file to use the 1mm nonlinearly generated template and
>> double the
>> numbers in the --subsamp option (i.e. --subsamp=4,4,2,2,1,1 becomes
>> --subsamp=8,8,4,4,2,2). This will give you what you want without
>> taking
>> an
>> inordinate amount of time.
>>
>> Hope this clarifies everything.
>>
>> Peace,
>>
>> Matt.
>>
>> -----Original Message-----
>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>> Behalf
>> Of Siew-Min Gan
>> Sent: Tuesday, May 19, 2009 4:50 PM
>> To: [log in to unmask]
>> Subject: Re: [FSL] choice of registration and MNI template
>>
>>> Hi Matt,
>> sorry I'm a bit unclear now. Do you mean there
>> might be
>> translational difference in my normalised FA if I use a
>> 2mm or 1mm MNI template ( with different FOV) for
>> registration? Is that due to the interpolation when the
>> normalised FA is resampled? If I'm doing the registration
>> FA2T1 and T12MNI, so I can get the inverse matrix to
>> transfer the template rois back to my FA native space,
>> would this translation still occur when the binary rois
>> are interpolated via nearest neighbour (i.e the rois
>> positioning would differ by translation when placed in
>> the native FA space depending on which MNI template I
>> use)?
>>
>> The fMRIB config file FA_2_FMRIB58_1mm.cnf uses the 1mm. Please
>> correct me
>> if I'm wrong.
>>
>> thanks
>>
>> Siewmin
>>
>> On the first point, I think you do, but am not 100% sure. The
>> differences
>>> would only be translational. For the nonlinear registration, the
>>> warp
>>> fields are on the order of 8-10mm in resolution so it never makes a
>>> difference in the warpfield to register at a higher resolution.
>>> What it
>>> does make a difference in is processing time and resource usage.
>>> Thus,
>>> in
>>> none of the fMRIB generated config files is a reference image used
>>> with
>>> less
>>> than a 2mm resolution (or a subsampling step that makes the image
>>> 2mm
>>> internally).
>>>
>>> Peace,
>>>
>>> Matt.
>>>
>>> -----Original Message-----
>>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>> Behalf
>>> Of Siew-Min Gan
>>> Sent: Tuesday, May 19, 2009 3:15 PM
>>> To: [log in to unmask]
>>> Subject: Re: [FSL] choice of registration and MNI template
>>>
>>>> Thanks for your reply.
>>> Can I clarify with you two points? You mentioned I need to take into
>>> account the FOVs in the applywarp step, do you just mean that my
>>> transformed FA in MNI space will take on the FOV of the MNI template
>>> that
>>> I'm registerering to?
>>> Secondly, my FA image is 2mm isotropic, my T1 image is 1mm
>>> isotropic.
>>> For
>>> nonlinear registration of my T1 image to the MNI template, do you
>>> mean
>>> it
>>> makes no difference for the quality of registration whether I
>>> choose the
>>> 1mm or the 2mm template MNI template? I.e it doesn't make better
>>> registration if the target image (MNI brain) is same resolution as
>>> source
>>> image (my T1)?
>>>
>>> Thanks
>>>
>>> Siewmin
>>>
>>> It really is fine to do the registration in two steps. You just
>>> don't
>>>> want
>>>> to resample your data (i.e. your ROIs) multiple times, so you can
>>>> use
>>>> the
>>>> options of applywarp to do your resampling in a single step. If,
>>>> for
>>>> some
>>>> reason, you wanted to do quantitative analysis on the FA in MNI
>>>> space,
>>>> it
>>>> would be better register the FA to the T1 and the T1 to the MNI and
>>>> then
>>>> use
>>>> apply warp to move the FA to MNI space in one step. Does this make
>>>> sense?
>>>>
>>>> I am not sure what the best way to deal with the different FOVs.
>>>> One
>>>> could
>>>> simply "register" the two templates together using 3 DOF
>>>> (translations
>>>> only), however maybe there is a more elegant way... (although voxel
>>>> coordinates are different, mm coordinates will be the same in both
>>>> templates). The FOVs will not make a difference for the quality
>>>> of the
>>>> registrations, but you might need to take them into account for the
>>>> applywarp step.
>>>>
>>>> Peace,
>>>>
>>>> Matt.
>>>>
>>>> -----Original Message-----
>>>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>>> Behalf
>>>> Of Siew-Min Gan
>>>> Sent: Monday, May 18, 2009 10:35 AM
>>>> To: [log in to unmask]
>>>> Subject: Re: [FSL] choice of registration and MNI template
>>>>
>>>>> Thanks.
>>>>
>>>> Sory I wasn't clear with the 3rd Question. The two ways mentioned
>>>> have
>>>> the
>>>> similiarity of first registering FA to T1 (linearly) for the same
>>>> subject.
>>>> The difference is with the 1st way:
>>>> after the linear registration of FA2T1, I then nonlinearly
>>>> register T1
>>>> from it's T1 native space directly to MNI standard space. Hence Fa
>>>> image
>>>> is moved twice to get into the standard space (Fa2T1 and T12MNI).
>>>> Likewise
>>>> the atlas rois from standard space is transformed back to the
>>>> native FA
>>>> image via two inverse matrices ( of FA2T1 and T12MNI)
>>>> With the second way
>>>> 2) after getting the linear regisration matrix of FA2T1, I apply
>>>> the
>>>> inverse of the registration matrix of FA2T1 on the T1 image,
>>>> moving T1
>>>> onto FA image space. I then nonlinearly register this registered
>>>> T1
>>>> (which is now in FA space) onto MNI standard space. This is
>>>> different
>>>> to
>>>> above where the T1 was registered from it's native space. Hence, in
>>>> contrast to above, the Fa image would only need to move once to get
>>>> into
>>>> the standard space via the nonlinear transformation matrix of the
>>>> registeredT1to MNI.Likewise the atlas rois from the standard
>>>> space is
>>>> transformed back to the native FA image via only one inverse
>>>> matrix (
>>>> of
>>>> the RegisteredT12MNI.
>>>>
>>>> I wonder which one would be more accurate to transform the roi
>>>> back to
>>>> the
>>>> native FA space ?
>>>>
>>>> With the choice of the MNI template, the ROIs of the atlas are
>>>> created
>>>> and
>>>> drawn on an average dti map which is normalised to the space of the
>>>> linearly created MNI152 182x218x182 1mm atlas. The MNI152 2mm atlas
>>>> used
>>>> in the provided cnf file is nonlinearly created and is of
>>>> dimension i.e
>>>> 91
>>>> x 109 x91 . If the dimensions and resolution are not important to
>>>> get
>>>> the
>>>> registration matrix to backtransform the standard space rois to
>>>> native
>>>> fa
>>>> image, but would there be some dissimiliarity between the linearly
>>>> created
>>>> and the nonlinearly created MNI152 template? Hence, I thought it
>>>> would
>>>> be
>>>> more "accurate" to register to the linearly created MNI template
>>>> which
>>>> is
>>>> used when creating the rois, although this would be at the
>>>> expense of
>>>> the
>>>> registration? Please correct me if I'm wrong.
>>>>
>>>> Thanks
>>>>
>>>> Siewmin
>>>>
>>>>
>>>>
>>>> You should use the default file in the T1_2_MNI152_2mm.cnf as
>>>> that will
>>>>> give
>>>>> the best results. That config file is properly tuned to give
>>>>> optimal
>>>>> T1
>>>>> to
>>>>> MNI template registrations. Why use a worse quality reference
>>>>> image
>>>>> (the
>>>>> linear template) with higher resolution (much longer processing
>>>>> time
>>>>> and
>>>>> higher resource usage for no benefit in registration quality)?
>>>>>
>>>>> Your commands look correct to me.
>>>>>
>>>>> 1. Yes
>>>>>
>>>>> 2. See above you shouldn't modify the config file.
>>>>>
>>>>> 3. I am having some difficulty following you. So long as you have
>>>>> transformations describing FA -> T1 and T1 -> MNI, you can move
>>>>> anything
>>>>> you
>>>>> want from FA to MNI or MNI to FA. Applywarp will only resample
>>>>> the
>>>>> images
>>>>> once, even if you include both a linear and nonlinear
>>>>> transformation
>>>>> (in
>>>>> fact you can include up to two linear transformations, one
>>>>> before and
>>>>> one
>>>>> after the nonlinear one) so long as you give everything in one
>>>>> commandline.
>>>>> You can also combine linear and nonlinear transformations with
>>>>> convertwarp.
>>>>>
>>>>>
>>>>> 4. Again I don't think you should be using the linearly derived
>>>>> templates.
>>>>>
>>>>> I'm not sure of the best answer to your last two questions.
>>>>>
>>>>> Peace,
>>>>>
>>>>> Matt.
>>>>>
>>>>> -----Original Message-----
>>>>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]]
>>>>> On
>>>>> Behalf
>>>>> Of Siewmin Gan
>>>>> Sent: Monday, May 18, 2009 8:00 AM
>>>>> To: [log in to unmask]
>>>>> Subject: [FSL] choice of registration and MNI template
>>>>>
>>>>> Hi, I have a few questions about trying fnirt to register fa and
>>>>> other
>>>>> scalar maps to the
>>>>> MNI template, so I can use the invert transformations to put the
>>>>> rois
>>>>> of
>>>>> the
>>>>> white matter
>>>>> atlas back onto the native fa and scalar maps. Apologies for the
>>>>> long
>>>>> questions.
>>>>>
>>>>> I performed linear 6DOF registration of subjects FA to their T1,
>>>>> and
>>>>> linear
>>>>> followed by
>>>>> non_linear registration of T1 to MNI152. The fa and scalar maps
>>>>> are
>>>>> calculated from 4D
>>>>> DWI with B0 unwarping/undistortion performed. These maps are 2mm
>>>>> isotropic
>>>>> and the
>>>>> T1 images of the subjects are 1mm isotropic. The MNI template
>>>>> chosen
>>>>> is
>>>>> the
>>>>> MNI_linear
>>>>> template 1mm. I did this following similiar steps to the 2nd fnirt
>>>>> example
>>>>> script on the
>>>>> fnirt website (i.e fmri to MNI via T1) and with slight
>>>>> modification of
>>>>> the
>>>>> T1_2_MNI152_2mm.cnf. May I ask if the following commands are the
>>>>> right
>>>>> way
>>>>> and
>>>>> quickest way to "concantenate" the two inverse linear T12FA
>>>>> matrix and
>>>>> nonlinear MNI2T1
>>>>> warp coefficient to transform binary rois from MNI to the native
>>>>> FA
>>>>> space?
>>>>> I have also
>>>>> listed my questions below about the choice of registration,
>>>>> template
>>>>> and
>>>>> using these
>>>>> appropriate parameters in the config file:
>>>>>
>>>>> T1_brain and Image_FA_brain (betted) The Image_FA I have is
>>>>> betted so
>>>>> I
>>>>> don't have a FA
>>>>> image with skull.
>>>>>
>>>>> flirt -ref T1_brain -in Image_FA_brain -out FA2T1_brain -omat
>>>>> FA2T1.mat;
>>>>> flirt -ref MNI152lin_T1_1mm_brain -in T1_brain -omat
>>>>> my_affine_transf.mat;
>>>>> fnirt --in=T1 --aff=my_affine_transf.mat --
>>>>> cout=my_nonlinear_transf --
>>>>> config=T1_2_MNI152lin_1mm.cnf;
>>>>> applywarp --ref=MNI152lin_T1_1mm --in=Image_FA_brain
>>>>> --warp=my_nonlinear_transf -
>>>>> -premat=FA2T1.mat --out=my_warped_fa2mni_1mm
>>>>>
>>>>> (applying inverse matrix to place ROI from MNI to FA native space)
>>>>> convert_xfm -omat T12FA.mat -inverse FA2T1.mat
>>>>> invwarp --ref=T1.nii.gz --warp=my_nonlinear_transf.nii.gz
>>>>> --out=nonlinear_MNI2T1
>>>>> applywarp --ref=Image_FA_brain --in=ROIs_in_MNI_space
>>>>> --warp=nonlinear_MNI2T1 --
>>>>> postmat=T12FA.mat out=ROIs_in_FAnative_space --interp=nn
>>>>>
>>>>>
>>>>> 1. Is it ok that I use a betted FA image all the way in these
>>>>> steps,
>>>>> as
>>>>> long
>>>>> as the T1
>>>>> image used in FNIRT is the original T1 with skull on?
>>>>>
>>>>> 2. I use the MNI152_lin_1mm template with slight modifications to
>>>>> T1_2_MNI152_2mm.cnf ( renaming it T1_2_MNI142lin_1mm.cnf ). The
>>>>> MNItemplate
>>>>> now
>>>>> chosen is a lot smoother, is 1mm and of different intensity to the
>>>>> other
>>>>> MNI
>>>>> template
>>>>> used in T1_2_MNI152_2mm.cnf). Apart from modifying the cnf file by
>>>>> changing
>>>>> the MNI
>>>>> template to the linear 1mm template, and the corresponding brain
>>>>> mask
>>>>> ,
>>>>> which other
>>>>> parameters would be important to change (my T1 and the
>>>>> MNIlin_1mm are
>>>>> both
>>>>> 1mm in
>>>>> resolution? Would there be any recommendations you suggest for the
>>>>> some
>>>>> of
>>>>> parameters in the config file in this circumstance: The current
>>>>> settings
>>>>> in
>>>>> the
>>>>> T1_2_MNI152_2mm config files are
>>>>>
>>>>> subsamp:4,4,2,2,1,1
>>>>> infwhm: 8,6,5,4.5,3,2
>>>>> refwhm:8,6,5,4,2,0
>>>>> lambda:300,150,100,50,40,30
>>>>> intorder:5
>>>>> biasres: 50 50 50
>>>>>
>>>>> 3. If I perform registration of image fa -->t1 -->to mni, without
>>>>> including
>>>>> the -out in the
>>>>> command line, the fa imagehas to be resampled once when nonlinear
>>>>> transformation to
>>>>> the MNI 1mm space is performed. Alternatively, I can use the
>>>>> inverse
>>>>> matrix
>>>>> of
>>>>> FA2T1.mat (i.e T12FA.mat) on T1. This will register T1 to FA
>>>>> followed
>>>>> by
>>>>> nonlinear
>>>>> transformation of this registeredT1 to the MNI template to get the
>>>>> my_nonlinear_transf
>>>>> matrix file of the T1(inFA native space) to MNI, which I can use
>>>>> to
>>>>> transform FA to MNI in
>>>>> one step. With the 1st method, the rois of the white matter
>>>>> template
>>>>> would
>>>>> be
>>>>> transformed onto the raw FA image using the inverse of
>>>>> my_nonlinear_transf
>>>>> matrix and
>>>>> FA2T1.mat as written in the command line above. With the second
>>>>> method,
>>>>> only
>>>>> the
>>>>> inverse of my_nonlinear_transf matrix would be used, without
>>>>> requiring
>>>>> the
>>>>> postmat
>>>>> T12FA.mat. With the Fa_image contrast and resolution, which way
>>>>> would
>>>>> be
>>>>> more
>>>>> precise/accurate to i) register or normalise ( register and
>>>>> resample
>>>>> FA
>>>>> images to MNI
>>>>> template) and ii) back-register the rois (by neighbouring
>>>>> interpolation)from
>>>>> the template
>>>>> to the raw space of the FA image?
>>>>>
>>>>> 4. The rois of the white matter atlas is created when
>>>>> normalising to
>>>>> the
>>>>> MNI152_lin_1mm
>>>>> template and not the MNI152 _1mm (the non linear template which
>>>>> has a
>>>>> higher
>>>>>
>>>>> resolution). If I want to invert transform the rois of the atlas
>>>>> in
>>>>> MNI
>>>>> space to the native
>>>>> space of the fa images, would it be right to use the
>>>>> MNI152_lin_1mm
>>>>> template
>>>>> to get the
>>>>> transformation matrices(because of how the rois of the atlas has
>>>>> been
>>>>> created), even
>>>>> though it is of poorer resolution than the other nonlinear MNI
>>>>> template
>>>>> ?
>>>>>
>>>>>
>>>>> 5. Is there any output from running the flirt and fnirt that can
>>>>> be
>>>>> used
>>>>> to
>>>>> get a measure
>>>>> of the precision in the registration methods (apart from visual
>>>>> inspections), or there a
>>>>> paper of fnirt that mentioned the precision of fnirt? I read
>>>>> that one
>>>>> way
>>>>> to
>>>>> quantitate the
>>>>> registration quality of the rois apart from visual inspection is
>>>>> assess
>>>>> the
>>>>> amount of
>>>>> displacement of x,y, z coordinated of defined landmarks from the
>>>>> MNI
>>>>> space
>>>>> when they
>>>>> are transferred to the normalised FA images?
>>>>>
>>>>> 6. Lastly, on the fnirt website, it mentioned that fnirt method
>>>>> is not
>>>>> diffeomorphic by
>>>>> consruction with some explainations of the difference. Would that
>>>>> matter
>>>>> in
>>>>> my case
>>>>> whether I use a diffeomorphic by construction method or not for
>>>>> the
>>>>> purpose
>>>>> I'm trying to
>>>>> achieve here(i.e to try as best to register binary rois from the
>>>>> atlas
>>>>> to
>>>>> the native fa
>>>>> space)?
>>>>>
>>>>> Many thanks for your kind patience.
>>>>>
>>>>> Siewmin
>>>>>
>>>>
>>>
>>
>
|