Re: [Freesurfer] T2pial/FLAIRpial processing issues

Allright; I digged further and further...., and finally found out that the sform of our T1 weighted image gets crushed during a lesion filling step earlier in the pipeline. I never recognized this while using FLIRT, maybe because FLIRT uses a different initialization (COG?) than using the header transformation as an initial registration...

Anyway, thanks for all the help. It still would be helpfull (I think) to have some kind of QC step in there which could warn if things got wrong.

Best, Martijn

On Fri, Nov 15, 2013 at 7:52 PM, Douglas N Greve greve@nmr.mgh.harvard.eduwrote:

Can you convert the two to mgz or nii with mri_convert and then run tkregister2? If they were acquired in the same session, then the geometric information in the header should bring them into close registration regardless of voxel size, orientation, field of view, etc. mri_convert should preserve this information. dcm2nii should do the same thing, but I did not write it so I don't know what it is doing:).

On 11/15/2013 01:15 PM, Martijn Steenwijk wrote:

...

Yes, they were acquired in the same session, but using a slightly different FOV. They were converted from dicom to nifti using dcm2nii and then used for FS analysis.

The axis orientation of the images is the same. Extensive image analysis (including registration of FLAIR and T1 using FLIRT) has been run on these images, but I never faced such a problem.

Is it possible that left/right orientation is is flipped by FLIRT due to the initial misalignment?

On Fri, Nov 15, 2013 at 6:45 PM, Douglas N Greve < greve@nmr.mgh.harvard.edu mailto:greve@nmr.mgh.harvard.edu> wrote:

were these acquired in the same session? If so, were they
converted to mgz directly from dicom?


On 11/15/2013 10:15 AM, Martijn Steenwijk wrote:

    I digged further into this.

    The scale difference previously reported is not true. It is
    only a translation error.

    It looks like that the orig.mgz has (0,0,0) in the center of
    the image; while FLAIRraw (after header-only registration) has
    the origin at index (0,0,0) - so in one of the corners of the
    volume. This means the initial (header based) transformation
    provided to flirt is also not correct.


    On Fri, Nov 15, 2013 at 12:49 PM, Martijn Steenwijk
    <martijnsteenwijk@gmail.com
    <mailto:martijnsteenwijk@gmail.com>
    <mailto:martijnsteenwijk@gmail.com

    <mailto:martijnsteenwijk@gmail.com>>> wrote:


        Dear all,

        Thanks for all the input. Doug, concerning your suggestion
        tkregister2 --mov FLAIRraw.mgz --targ orig.mgz --regheader
    --reg junk

        Using this, the axes orientationseem to be correct, but
    scaling
        and translation of registered FLAIRraw seem to be way off.
        (registered FLAIRraw is much larger compared to orig.mgz
    (in the
        order of multiple times) and is translated to the
    radiologically
        left-caudal part of orig.mgz.

        mri_convert -rl gives a similar result as using tkregister2 as
        proposed by Doug.

        Best,
        Martijn


        On Thu, Nov 14, 2013 at 8:38 PM, Douglas N Greve
        <greve@nmr.mgh.harvard.edu
    <mailto:greve@nmr.mgh.harvard.edu>
    <mailto:greve@nmr.mgh.harvard.edu

    <mailto:greve@nmr.mgh.harvard.edu>>> wrote:


            It should not make a difference. We compute an initial
            registration
            based on the header and pass that to FLIRT. Can you verify
            that this
            initial registration is correct? You can do that with

            tkregister2 --mov FLAIRraw.mgz --targ orig.mgz --regheader
            --reg junk

            If that looks ok, then it is possible that flirt is
    just failing.

            Also, newer versions of FS will use an internal version of
            flirt and so
            not reliant on FSLDIR being set or FSL being installed.

            doug




            On 11/14/2013 12:52 PM, Bruce Fischl wrote:
            > Hi Matt
            >
            > hopefully Doug will chime in, but I don't see why
    that is
            the case. We
            > preserve all the RAS information when we "conform",
    so the
            info should
            > be there for flirt
            >
            > cheers
            > Bruce
            >
            >
            > On Thu, 14 Nov 2013, Matt Glasser wrote:
            >
            >> I think you can do it with mri_convert ?rl.  I
    don't think
            it makes
            >> sense to
            >> expect FLIRT to get the registration right reliably
    if the
            axes are
            >> off (in
            >> the HCP Pipelines we definitely don't expect this
    and make
            sure all
            >> images
            >> are RPI before processing them).
            >>
            >> Peace,
            >>
            >> Matt.
            >>
            >> From: Martijn Steenwijk <martijnsteenwijk@gmail.com
    <mailto:martijnsteenwijk@gmail.com>
            <mailto:martijnsteenwijk@gmail.com

    <mailto:martijnsteenwijk@gmail.com>>>

            >> Date: Thursday, November 14, 2013 9:27 AM
            >> To: Bruce Fischl <fischl@nmr.mgh.harvard.edu
    <mailto:fischl@nmr.mgh.harvard.edu>
            <mailto:fischl@nmr.mgh.harvard.edu

    <mailto:fischl@nmr.mgh.harvard.edu>>>
            >> Cc: Veronica Popescu <v.popescu@vumc.nl
    <mailto:v.popescu@vumc.nl>
            <mailto:v.popescu@vumc.nl

    <mailto:v.popescu@vumc.nl>>>, Hugo Vrenken
            >> <h.vrenken@vumc.nl <mailto:h.vrenken@vumc.nl>
    <mailto:h.vrenken@vumc.nl <mailto:h.vrenken@vumc.nl>>>,
            >> freesurfer <freesurfer@nmr.mgh.harvard.edu
    <mailto:freesurfer@nmr.mgh.harvard.edu>
            <mailto:freesurfer@nmr.mgh.harvard.edu

    <mailto:freesurfer@nmr.mgh.harvard.edu>>>

            >> Subject: Re: [Freesurfer] T2pial/FLAIRpial
    processing issues
            >>
            >> Hi Matt, Bruce,
            >> The problems are indeed a flirt issue, but given
    that it's
            programmed
            >> with
            >> --init-fsl there is not much flexibility.
            >>
            >> However, from a FS point of view, it might be a
    better and
            more robust
            >> approach to first register FLAIRraw.mgz to raw.mgz,
    and then
            >> concatenate the
            >> resulting registration with the (header based)
    registration
            from
            >> raw.mgz to
            >> orig.mgz - instead of trusting the robustness of either
            registration
            >> algorithms.
            >>
            >> ... or, as Matt suggests, fix the image axes
    appropriately
            after
            >> converting
            >> the FLAIR to mgz, and then run the registration.
    But I'm
            not sure how
            >> to do
            >> that.
            >>
            >> Best,
            >> Martijn
            >>
            >>
            >> On Thu, Nov 14, 2013 at 5:53 PM, Bruce Fischl
            >> <fischl@nmr.mgh.harvard.edu
    <mailto:fischl@nmr.mgh.harvard.edu>
            <mailto:fischl@nmr.mgh.harvard.edu

    <mailto:fischl@nmr.mgh.harvard.edu>>>

            >> wrote:
            >>       Hi Martijn
            >>
            >>       we can print warning and errors if FSLDIR is
    not set,
            but the
            >>       registration errros seem to be mostly a flirt
    issue,
            no? Have
            >>       you posted on the FSL list?
            >>
            >>       Bruce
            >>
            >>       On Thu, 14 Nov 2013, Martijn Steenwijk wrote:
            >>
            >>             Dear all,
            >>             There seem to be some serious issues
    with the
            T2pial
            >>             / FLAIRpial processing
            >>             options provided in the latest versions of
            >>             FreeSurfer.
            >>
            >>             First of all, although not clearly
    documented,
            >>             T2pial / FLAIRpial processing
            >>             does require the FSLDIR to be set (in
    order to use
            >>             bbregister with
            >>             --init-fsl). If not set,  recon-all will
            >>             finish just without using the T2 or
            >>             FLAIR information. Importantly, no
    clear warning
            >>             about this is given
            >>             although the user expects that T2/FLAIR
    has been
            >>             used for pial surface
            >>             refinement. I think this is a very easy
    source of
            >>             errors, so it might be a
            >>             good idea to just throw an error when
    FSLDIR is
            >>             missing and T2pial/FLAIRpial
            >>             processing is requested.
            >>
            >>             Second, if FSLDIR has been set.
    T2pial/FLAIRpial
            >>             uses bbregister with FLIRT
            >>             initialisation to align the T2 or FLAIR
    image to
            >>             orig.mgz. However, we have
            >>             some high-res 3DFLAIR data in which this
            >>             registration (more specifially, the
            >>             FLIRT initialisation step of
    bbregister) seems to
            >>             fail in more than half of
            >>             the cases. Apparently, the FSL
    initialisation
            is not
            >>             capable to change the
            >>             orientation such that it fits to the
    coordinate
            >>             system used for orig.mgz.
            >>             Again, no warning or error is thrown,
    but the
            >>             processing just continues with
            >>             the wrong registration and without
    noticing the
            >>             results will get worse
            >>             compared to not using T2 or FLAIR.
            >>             Although I know its essential to look
    at the output
            >>             data; would it be
            >>             possible to put some effort in letting
    the user
            know
            >>             that things got wrong?
            >>
            >>
            >>             As a sidenote; I tried to fix this
    registration
            >>             issue, but it seems to be
            >>             very complicated using bbregister
    (other than just
            >>             inserting my own
            >>             registration obtained by using FLIRT to
    register
            >>             native FLAIR to native T1,
            >>             and subsequently transform the result
    to Freesurfer
            >>             space in order to obtain
            >>             FLAIR.mgz).
            >>
            >>             Best,
            >>             Martijn
            >>
            >>
            >>
            >>
            >> The information in this e-mail is intended only for the
            person to whom
            >> it is
            >> addressed. If you believe this e-mail was sent to
    you in
            error and the
            >> e-mail
            >> contains patient information, please contact the
    Partners
            Compliance
            >> HelpLine at
            >> http://www.partners.org/complianceline . If the
    e-mail was
            sent to you
            >> in error
            >> but does not contain patient information, please
    contact
            the sender
            >> and properly
            >> dispose of the e-mail.
            >>
            >>
            >> _______________________________________________
    Freesurfer
            mailing list
            >> Freesurfer@nmr.mgh.harvard.edu
    <mailto:Freesurfer@nmr.mgh.harvard.edu>
            <mailto:Freesurfer@nmr.mgh.harvard.edu

    <mailto:Freesurfer@nmr.mgh.harvard.edu>>

            >>
    https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The
            >> information
            >> in this e-mail is intended only for the person to
    whom it is
            >> addressed. If
            >> you believe this e-mail was sent to you in error
    and the
            e-mail contains
            >> patient information, please contact the Partners
    Compliance
            HelpLine at
            >> http://www.partners.org/complianceline . If the
    e-mail was
            sent to
            >> you in
            >> error but does not contain patient information, please
            contact the
            >> sender
            >> and properly dispose of the e-mail.
            >>
            >
            >
            > _______________________________________________
            > Freesurfer mailing list
            > Freesurfer@nmr.mgh.harvard.edu
    <mailto:Freesurfer@nmr.mgh.harvard.edu>
            <mailto:Freesurfer@nmr.mgh.harvard.edu

    <mailto:Freesurfer@nmr.mgh.harvard.edu>>

            >
    https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

            --
            Douglas N. Greve, Ph.D.
            MGH-NMR Center
    greve@nmr.mgh.harvard.edu <mailto:greve@nmr.mgh.harvard.edu>
    <mailto:greve@nmr.mgh.harvard.edu

    <mailto:greve@nmr.mgh.harvard.edu>>
            Phone Number: 617-724-2358 <tel:617-724-2358>
    <tel:617-724-2358 <tel:617-724-2358>>
            Fax: 617-726-7422 <tel:617-726-7422> <tel:617-726-7422

    <tel:617-726-7422>>

            Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
    <http://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting>
            <http://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting>
            FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2
    www.nmr.mgh.harvard.edu/facility/filedrop/index.html
    <http://www.nmr.mgh.harvard.edu/facility/filedrop/index.html>
                   <http://www.nmr.mgh.harvard.

edu/facility/filedrop/index.html>

            Outgoing:
    ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/

            _______________________________________________
            Freesurfer mailing list
    Freesurfer@nmr.mgh.harvard.edu
    <mailto:Freesurfer@nmr.mgh.harvard.edu>
            <mailto:Freesurfer@nmr.mgh.harvard.edu

    <mailto:Freesurfer@nmr.mgh.harvard.edu>>
    https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer




--     Douglas N. Greve, Ph.D.
MGH-NMR Center
greve@nmr.mgh.harvard.edu <mailto:greve@nmr.mgh.harvard.edu>
Phone Number: 617-724-2358 <tel:617-724-2358>
Fax: 617-726-7422 <tel:617-726-7422>

Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
<http://surfer.nmr.mgh.harvard.edu/fswiki/BugReporting>
FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2
www.nmr.mgh.harvard.edu/facility/filedrop/index.html
<http://www.nmr.mgh.harvard.edu/facility/filedrop/index.html>
Outgoing:
ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/

-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422

Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/