Re: [Freesurfer] possible to use freesurfer segmentations to get mean values for anatomic structures from other sequences

On 5/26/13 2:11 PM, Salil Soman wrote:

Dear Doug,

Thank you for your email. Following up on our previous discussions:

1. *bbregister fail *- this was happening because of error from

fslregister. The variable fsvol was translating to "ERROR". In fslregister th value for fsvol is set to brainmask, along with numerous other parameters being initialized. By adding the line "set fsvol = brain mask" (without quotes) at the beginning of the section "#Convert reference to analyze" the CentOS 6 version I am running on ubuntu 12.04 (with SGE), now works.

see previous msg.

2. when you suggested I run bbregister on the lowb.nii file, was it

because I had suggested I wanted to register to a 2D ADC map? If I want to instead use a 3D volume (e.g. a 3D ASL), would I use the actual volume I want to get value from instead of lowb.nii? I ended up running

Oh, I did not realize it was a 2D map. Why do you have a 2D map instead of 3D? That might be hard to get registered. You could have used the ADC or FA. In general, you don't want to use the contrast that you will later be testing as it creates a little bit of circularity. It is not much though, and it should be fine.

3. for the mri_vol2vol step - I had to change targ to fstarg, and I

left out the interp flag so it would do its default trilinear: mri_vol2vol --mov Vol_906.nii --fstarg T1.mgz --o Vol_906_Reg_Bold.nii --no-save-reg --reg register_bold.dat --inv

that maps T1 into the bold space. Is that what you want? If you want to map the bold into the anatomical space, then leave off -inv. This will also solve your problem below doug

4. I would like to get the segmentations from aparc+aseg.mgz as the

areas where I get the values summed. So I entered mri_segstats --seg mri/aparc+aseg.mgz --ctab $FREESURFER_HOME/FreeSurferColorLUT.txt --i Vol_906_Reg_Bold.nii --sum Vol_aparc_aseg.stats

but then I received this error:

ERROR: dimension mismatch between input volume and seg input 224 224 170 seg 256 256 256

*So, in summary, while I've fixed bbregister to work on my system, I am still having trouble normalizing my 3D non T1 volume data into the FS space to be able to use the labels from aparc+aseg.mgz to get statistics for each anatomic area.*

Any ideas on how to proceed?

Thanks,

Sal

On Sat, May 25, 2013 at 11:51 AM, Douglas Greve <greve@nmr.mgh.harvard.edu mailto:greve@nmr.mgh.harvard.edu> wrote:

On 5/25/13 1:39 PM, Salil Soman wrote:

...

Thank you Doug.

 Just so I am clear - is the anatomical the nifti T1.mgz or is it
a different file. From your email I gather should do the following:

*1)  Use bbregister to register the lowb image to the anatomical.
This creates a registration matrix.*

bbregister -s SUBJECTNAME --mov lowb.nii --initfsl --reg register.dat

That is right, but add --t2 (since it is t2 weighted).

...

*2) Use mri_vol2vol and the registration to map the ADC map to
the anatomical*
mri_vol2vol --mov lowb.nii --targ $vol --inv --interp nearest --o $vol2diff --reg $outdir/register.dat --no-save-reg

Use the ADC as the moveable (not lowb, but use the lowb for
bbregister). The target volume should be the T1.mgz (or any
conformed volume). The output will be the adc in the anatomical
space (not sure why you call it vol2diff).

...

*3) use mri_segstats to extract the values*
*mri_segstats*  --seg $SUBJECTS_DIR/SUBJECTNAME/mri/wmparc.mgz --ctab $FREESURFER_HOME/FreeSurferColorLUT.txt --i lowb.nii --sum fa.stats

The input would be the adc volume mapped the anatomical space. I
would use WMParcStatsLUT.txt or else you'll get a billion
different areas not represented in wmparc. You may also want to
add "--seg-erode 1" to erode the segmentations by a voxel. This
helps to prevent partial voluming.

doug

...

Where lowb.nii is the other MRI tissue sequence I am analyzing
(e.g. ADC), $vol2diff is the output image of the registration,
and fa.stats will be my output stats table?

I suspect there is part of the syntax for these tools I do not
understand. Also, what input would $vol be?

Best wishes,

Sal


On Sat, May 25, 2013 at 10:06 AM, Douglas Greve
<greve@nmr.mgh.harvard.edu <mailto:greve@nmr.mgh.harvard.edu>> wrote:


    Hi Sal, yes you can. Use bbregister to register the lowb
    image to the anatomical. This creates a registration matrix.
    Use mri_vol2vol and the registration to map the ADC map to
    the anatomical, then use mri_segstats to extract the values
    doug

    ps. Please post questions to the FS list instead of us
    directly so that others can benefit and it gets archived.
    thanks!



    On 5/25/13 1:03 PM, Salil Soman wrote:

        Dear Dr. Greve,

        Is it possible to register a nonstructural MR sequences
        (e.g. an ADC map) with the aseg.mgz file (or
        aparc+aseg.mgz file, etc) and then, using the
        segmentation from the aseg (or aparc+aseg) file to output
        a mean ADC value for each anatomic area segmented?

        Thank you for your time and consideration.

        Best wishes,

        Sal

        Salil Soman, MD, MS




    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.