External Email - Use Caution

Dear Dr Bruce, Thank you very much for your guidance. I ran recon-all on a subject "test" using Freesurfer V6.0 and the following command line is reported in the "recon-all.log":

/usr/local/freesurfer/6.0/bin/recon-all -all -qcache -i T1.nii.gz -flair FLAIR.nii.gz -s test

When I search for normalization of flair, I only find how Freesurfer normalized T1. I searched for "nonmax_suppress" and it is not in the log file. I am just wondering what I am doing wrong. Thanks again for any highlights about the correct method to normalize FLAIR

Sincerely,

Hi JamesB

there is T2/FLAIR normalization in recon-all if you process it through there. You can look for "nonmax_suppress" if you want to find the command line

cheers Bruce

On Fri, Sep 23, 2022 at 2:06 PM James Brown jb1979000@gmail.com wrote:

Dear Freesurfer experts, We appreciate any advice or guidance from your community of professionals and experts in the field about the issue below.

Thank you for your kindness.

Dear Freesurfer experts, In order to normalize white matter intensity of T1 images and correct for bias field. We do the following steps: mri_convert T1.nii.gz T1.mgz mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2 mri_normalize -g 1 T1_N3.mgzT1_norm.mgz mri_convert T1_norm.mgz T1_norm.nii.gz

We would like to apply bias field correction on FLAIR images then normalize white matter intensity. We tried N4biasfieldcorrection from ANTS but it didn't work as expected. Reviewing documents about N4biasfieldcorrection we realize that the algorithm has been developed to correct bias field correction in T1 not clear if it works on FLAIR images.

Are the steps above valid for normalizing FLAIR images or can it only be applied on T1 images? I would like to send a "recon-all.log" file that includes the processing steps of FLAIR. Is there any link I can use to upload the file. It gets rejected every time I send it due to size limitations.

We appreciate any feedback about the proper way to correct FLAIR images for bias fields and normalize white matter intensity in FLAIR images.

Thank you JamesB

On Wed, Sep 21, 2022 at 10:18 AM James Brown jb1979000@gmail.com wrote:

...

Dear Freesurfer experts, In order to normalize white matter intensity of T1 images and correct for bias field. We do the following steps: mri_convert T1.nii.gz T1.mgz mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2 mri_normalize -g 1 T1_N3.mgzT1_norm.mgz mri_convert T1_norm.mgz T1_norm.nii.gz

We would like to apply bias field correction on FLAIR images then normalize white matter intensity. We tried N4biasfieldcorrection from ANTS but it didn't work as expected. Reviewing documents about N4biasfieldcorrection we realize that the algorithm has been developed to correct bias field correction in T1 not clear if it works on FLAIR images.

Are the steps above valid for normalizing FLAIR images or can it only be applied on T1 images? I would like to send a "recon-all.log" file that includes the processing steps of FLAIR. Is there any link I can use to upload the file. It gets rejected every time I send it due to size limitations.

We appreciate any feedback about the proper way to correct FLAIR images for bias fields and normalize white matter intensity in FLAIR images.

Thank you JamesB

it looks like you are using v6. In v6, the flair might not have been normalized

On 9/23/2022 3:53 PM, Fischl, Bruce R.,PHD wrote:

I think you need to add something like -FLAIRpial or -T2pial so it uses it (in which case it will first normalize it), but hopefully someone else can confirm

cheers Bruce

---

*From:* freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu on behalf of James Brown jb1979000@gmail.com *Sent:* Friday, September 23, 2022 2:40 PM *To:* freesurfer@nmr.mgh.harvard.edu freesurfer@nmr.mgh.harvard.edu *Subject:* Re: [Freesurfer] White matter normalization of FLAIR images

External Email - Use Caution

Dear Dr Bruce, Thank you very much for your guidance. I ran recon-all on a subject "test" using Freesurfer V6.0 and the following command line is reported in the "recon-all.log":

/usr/local/freesurfer/6.0/bin/recon-all -all -qcache -i T1.nii.gz -flair FLAIR.nii.gz -s test

When I search for normalization of flair, I only find how Freesurfer normalized T1. I searched for "nonmax_suppress" and it is not in the log file. I am just wondering what I am doing wrong. Thanks again for any highlights about the correct method to normalize FLAIR

Sincerely,

Hi JamesB

there is T2/FLAIR normalization in recon-all if you process it through there. You can look for "nonmax_suppress" if you want to find the command line

cheers Bruce

On Fri, Sep 23, 2022 at 2:06 PM James Brown jb1979000@gmail.com wrote:

Dear Freesurfer experts,
We appreciate any advice or guidance from your community of
professionals and experts in the field about the issue below.

Thank you for your kindness.

Dear Freesurfer experts,
In order to normalize white matter intensity of T1 images and
correct for bias field. We do the following steps:
mri_convert T1.nii.gz T1.mgz
mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2
mri_normalize -g 1 T1_N3.mgzT1_norm.mgz
mri_convert T1_norm.mgz T1_norm.nii.gz

We would like to apply bias field correction on FLAIR images then
normalize white matter intensity.
We tried N4biasfieldcorrection from ANTS but it didn't work as
expected. Reviewing documents about N4biasfieldcorrection we
realize that the algorithm has been developed to correct bias
field correction in T1 not clear if it works on FLAIR images.

Are the steps above valid for normalizing FLAIR images or can it
only be applied on T1 images? I would like to send a
"recon-all.log" file that includes the processing steps of FLAIR.
Is there any link I can use to upload the file. It gets rejected
every time I send it due to size limitations.

We appreciate any feedback about the proper way to correct FLAIR
images for bias fields and normalize white matter intensity in
FLAIR images.

Thank you
JamesB

On Wed, Sep 21, 2022 at 10:18 AM James Brown <jb1979000@gmail.com>
wrote:

    Dear Freesurfer experts,
    In order to normalize white matter intensity of T1 images and
    correct for bias field. We do the following steps:
    mri_convert T1.nii.gz T1.mgz
    mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2
    mri_normalize -g 1 T1_N3.mgzT1_norm.mgz
    mri_convert T1_norm.mgz T1_norm.nii.gz

    We would like to apply bias field correction on FLAIR images
    then normalize white matter intensity.
    We tried N4biasfieldcorrection from ANTS but it didn't work as
    expected. Reviewing documents about N4biasfieldcorrection we
    realize that the algorithm has been developed to correct bias
    field correction in T1 not clear if it works on FLAIR images.

    Are the steps above valid for normalizing FLAIR images or can
    it only be applied on T1 images? I would like to send a
    "recon-all.log" file that includes the processing steps of
    FLAIR. Is there any link I can use to upload the file. It gets
    rejected every time I send it due to size limitations.

    We appreciate any feedback about the proper way to correct
    FLAIR images for bias fields and normalize white matter
    intensity in FLAIR images.

    Thank you
    JamesB

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer

1. Sigma is the smoothing kernel applied to the bias field so you can make it pretty large depending on how high frequency your bias field is (typically defined by the size of your receive coils) 2. min_dist specifies the minimum distance from the surface a voxel must be for it to be considered as a potential control point 3. I don’t see why not although I’ve never tried it. It won’t fix slice-specific artifacts in e.g. interlaced acquisitions 4. We estimate the bias field from the white matter, then apply it to the whole brain.

I don’t think nonmax_suppress will do anything without surfaces though

Cheers Bruce

From: freesurfer-bounces@nmr.mgh.harvard.edu freesurfer-bounces@nmr.mgh.harvard.edu On Behalf Of James Brown Sent: Tuesday, September 27, 2022 7:20 AM To: freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] White matter normalization of FLAIR images

External Email - Use Caution Dear Dr Greeve and Dr Fiscle, I am grateful to you guiding me through this. Indeed, I can see in recon-all.log how Freesurfer 7.3.2 is normalizing the Flair images. it uses the command below:

<mri_normalize -seed 1234 -sigma 0.5 -nonmax_suppress 0 -min_dist 1 -aseg aseg.presurf.mgz -surface lh.white identity.nofile -surface rh.white identity.nofile FLAIR.prenorm.mgz FLAIR.norm.mgz >

I tried to run the previous command without the flag -surface" and it seems to be working. using this command

<mri_normalize -seed 1234 -sigma 0.5 -nonmax_supress 0 -min_dist 1 --no-skull >

I would like to inquire about the following points and I highly appreciate any clarification you might provide: 1- It seems mri_normalize uses default sigma of 8 for normalizing T1 images and 0.5 for Flair. How far can we go up with this value? More specifically, when I have noticed when I increase sigma in noisy flair images in my dataset, I get better out put images in terms of less noticeable motion artifacts. Is there any cut off value for sigma? Can I use high values up to (e.g. 25)? 2- What does the flag "-min_dist" do? I tried to find information in mri_normalize man (i.e. help ) but I couldn't find information about this flag. I assume it has to do with the flag surface and they both should be used together? 3- Can I use mri_normalize on 2D flair images ? 4- Is it recommended to do whole brain normalization (white/gray and csf) or just white matter normalization within a mask.

Thank you very much for all that you do.

James

it looks like you are using v6. In v6, the flair might not have been normalized  On 9/23/2022 3:53 PM, Fischl, Bruce R.,PHD wrote: I think you need to add something like -FLAIRpial or -T2pial so it uses it (in which case it will first normalize it), but hopefully someone else can confirm

cheers Bruce

On Fri, Sep 23, 2022 at 2:40 PM James Brown <jb1979000@gmail.commailto:jb1979000@gmail.com> wrote: Dear Dr Bruce, Thank you very much for your guidance. I ran recon-all on a subject "test" using Freesurfer V6.0 and the following command line is reported in the "recon-all.log":

/usr/local/freesurfer/6.0/bin/recon-all -all -qcache -i T1.nii.gz -flair FLAIR.nii.gz -s test

When I search for normalization of flair, I only find how Freesurfer normalized T1. I searched for "nonmax_suppress" and it is not in the log file. I am just wondering what I am doing wrong. Thanks again for any highlights about the correct method to normalize FLAIR

Sincerely,

Hi JamesB

there is T2/FLAIR normalization in recon-all if you process it through there. You can look for "nonmax_suppress" if you want to find the command line

cheers Bruce

On Fri, Sep 23, 2022 at 2:06 PM James Brown <jb1979000@gmail.commailto:jb1979000@gmail.com> wrote: Dear Freesurfer experts, We appreciate any advice or guidance from your community of professionals and experts in the field about the issue below.

Thank you for your kindness.

Dear Freesurfer experts, In order to normalize white matter intensity of T1 images and correct for bias field. We do the following steps: mri_convert T1.nii.gz T1.mgz mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2 mri_normalize -g 1 T1_N3.mgzT1_norm.mgz mri_convert T1_norm.mgz T1_norm.nii.gz

We would like to apply bias field correction on FLAIR images then normalize white matter intensity. We tried N4biasfieldcorrection from ANTS but it didn't work as expected. Reviewing documents about N4biasfieldcorrection we realize that the algorithm has been developed to correct bias field correction in T1 not clear if it works on FLAIR images.

Are the steps above valid for normalizing FLAIR images or can it only be applied on T1 images? I would like to send a "recon-all.log" file that includes the processing steps of FLAIR. Is there any link I can use to upload the file. It gets rejected every time I send it due to size limitations.

We appreciate any feedback about the proper way to correct FLAIR images for bias fields and normalize white matter intensity in FLAIR images.

Thank you JamesB

On Wed, Sep 21, 2022 at 10:18 AM James Brown <jb1979000@gmail.commailto:jb1979000@gmail.com> wrote: Dear Freesurfer experts, In order to normalize white matter intensity of T1 images and correct for bias field. We do the following steps: mri_convert T1.nii.gz T1.mgz mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2 mri_normalize -g 1 T1_N3.mgzT1_norm.mgz mri_convert T1_norm.mgz T1_norm.nii.gz

We would like to apply bias field correction on FLAIR images then normalize white matter intensity. We tried N4biasfieldcorrection from ANTS but it didn't work as expected. Reviewing documents about N4biasfieldcorrection we realize that the algorithm has been developed to correct bias field correction in T1 not clear if it works on FLAIR images.

Are the steps above valid for normalizing FLAIR images or can it only be applied on T1 images? I would like to send a "recon-all.log" file that includes the processing steps of FLAIR. Is there any link I can use to upload the file. It gets rejected every time I send it due to size limitations.

We appreciate any feedback about the proper way to correct FLAIR images for bias fields and normalize white matter intensity in FLAIR images.

Thank you JamesB

If you just want to normalize the flair, you can run samseg on it, ie, samseg --i flair.mgz --o samseg.flair in the output folder you will find a bias corrected volume

On 9/28/2022 7:42 AM, James Brown wrote:

External Email - Use Caution

Dear Dr Bruce, Thank you very much for answering my questions and for help. We highly appreciate you.

All the best James

1. Sigma is the smoothing kernel applied to the bias field so you can make it pretty large depending on how high frequency your bias field is (typically defined by the size of your receive coils)
2. min_dist specifies the minimum distance from the surface a voxel must be for it to be considered as a potential control point
3. I don’t see why not although I’ve never tried it. It won’t fix slice-specific artifacts in e.g. interlaced acquisitions
4. We estimate the bias field from the white matter, then apply it to the whole brain.

I don’t think nonmax_suppress will do anything without surfaces though

Cheers

Bruce

On Tue, Sep 27, 2022 at 7:20 AM James Brown jb1979000@gmail.com wrote:

Dear Dr Greeve and Dr Fiscle,
I am grateful to you guiding me through this. Indeed, I can see in
recon-all.log how Freesurfer 7.3.2 is normalizing the Flair
images. /it uses the command below:/

<mri_normalize -seed 1234 -sigma 0.5 -nonmax_suppress 0 -min_dist
1 -aseg aseg.presurf.mgz -surface lh.white identity.nofile
-surface rh.white identity.nofile FLAIR.prenorm.mgz FLAIR.norm.mgz >

I tried to run the previous command without the flag -surface" and
it seems to be working. using this command

<mri_normalize -seed 1234 -sigma 0.5 -nonmax_supress 0 -min_dist 1
--no-skull >

I would like to inquire about the following points and I highly
appreciate any clarification you might provide:
1- It seems mri_normalize uses default sigma of 8 for normalizing
T1 images and 0.5 for Flair. How far can we go up with this
value? More specifically, when I have noticed when I increase
sigma in noisy flair images in my dataset, I get better out
put images in terms of less noticeable motion artifacts. Is there
any cut off value for sigma? Can I use high values up to (e.g. 25)?
2-  What does the flag "-min_dist" do? I tried to find information
in mri_normalize man (i.e. help ) but I couldn't find information
about this flag. I assume it has to do with the flag surface and
they both should be used together?
3- Can I use mri_normalize on 2D flair images ?
4- Is it recommended to do whole brain normalization (white/gray
and csf) or just white matter normalization within a mask.

Thank you very much for all that you do.

James





it looks like you are using v6. In v6, the flair might not have
been normalized

On 9/23/2022 3:53 PM, Fischl, Bruce R.,PHD wrote:

...

I think you need to add something like -FLAIRpial or -T2pial so
it uses it (in which case it will first normalize it), but
hopefully someone else can confirm

cheers
Bruce

On Fri, Sep 23, 2022 at 2:40 PM James Brown <jb1979000@gmail.com>
wrote:

    Dear Dr Bruce,
    Thank you very much for your guidance.
    I ran recon-all on a subject "test" using Freesurfer V6.0 and
    the following command line is reported in the "recon-all.log":

    /usr/local/freesurfer/6.0/bin/recon-all -all -qcache -i
    T1.nii.gz -flair FLAIR.nii.gz -s test

    When I search for normalization of flair, I only find how
    Freesurfer normalized T1.
    I searched for "nonmax_suppress" and it is not in the log
    file. I am just wondering what I am doing wrong. Thanks again
    for any highlights about the correct method to normalize FLAIR

    Sincerely,



    Hi JamesB

    there is T2/FLAIR normalization in recon-all if you process it
    through there. You can look for "nonmax_suppress" if you want
    to find the command line

    cheers
    Bruce

    On Fri, Sep 23, 2022 at 2:06 PM James Brown
    <jb1979000@gmail.com> wrote:

        Dear Freesurfer experts,
        We appreciate any advice or guidance from your community
        of professionals and experts in the field about the issue
        below.

        Thank you for your kindness.

        Dear Freesurfer experts,
        In order to normalize white matter intensity of T1 images
        and correct for bias field. We do the following steps:
        mri_convert T1.nii.gz T1.mgz
        mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2
        mri_normalize -g 1 T1_N3.mgzT1_norm.mgz
        mri_convert T1_norm.mgz T1_norm.nii.gz

        We would like to apply bias field correction on FLAIR
        images then normalize white matter intensity.
        We tried N4biasfieldcorrection from ANTS but it didn't
        work as expected. Reviewing documents about
        N4biasfieldcorrection we realize that the algorithm has
        been developed to correct bias field correction in T1 not
        clear if it works on FLAIR images.

        Are the steps above valid for normalizing FLAIR images or
        can it only be applied on T1 images? I would like to send
        a "recon-all.log" file that includes the processing steps
        of FLAIR. Is there any link I can use to upload the file.
        It gets rejected every time I send it due to size
        limitations.

        We appreciate any feedback about the proper way to correct
        FLAIR images for bias fields and normalize white matter
        intensity in FLAIR images.

        Thank you
        JamesB

        On Wed, Sep 21, 2022 at 10:18 AM James Brown
        <jb1979000@gmail.com> wrote:

            Dear Freesurfer experts,
            In order to normalize white matter intensity of T1
            images and correct for bias field. We do the following
            steps:
            mri_convert T1.nii.gz T1.mgz
            mri_nu_correct.mni --i T1.mgz --o T1_N3.mgz --n 2
            mri_normalize -g 1 T1_N3.mgzT1_norm.mgz
            mri_convert T1_norm.mgz T1_norm.nii.gz

            We would like to apply bias field correction on FLAIR
            images then normalize white matter intensity.
            We tried N4biasfieldcorrection from ANTS but it didn't
            work as expected. Reviewing documents about
            N4biasfieldcorrection we realize that the
            algorithm has been developed to correct bias field
            correction in T1 not clear if it works on FLAIR images.

            Are the steps above valid for normalizing FLAIR images
            or can it only be applied on T1 images? I would like
            to send a "recon-all.log" file that includes the
            processing steps of FLAIR. Is there any link I can use
            to upload the file. It gets rejected every time I send
            it due to size limitations.

            We appreciate any feedback about the proper way to
            correct FLAIR images for bias fields and normalize
            white matter intensity in FLAIR images.

            Thank you
            JamesB

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer