External Email - Use Caution
All,
I've tried the dev version with FS_HISTO_STD_THRESH set to 20, but I am seeing the same error message.
hayashis@xps15-9560:~/test/deface/app-deface(master*) $ export FS_HISTO_STD_THRESH=20 hayashis@xps15-9560:~/test/deface/app-deface(master*) $ ./mri_deface ../t1.nii.gz $FREESURFER_HOME/average/talairach_mixed_with_skull.gca $FREESURFER_HOME/average/face.gca t1.nii.gz logging results to t1.nii.log reading '/home/hayashis/app/freesurfer-dev/average/talairach_mixed_with_skull.gca'... reading '../t1.nii.gz'... changing type of input volume to 8 bits/voxel... MRIchangeType: Building histogram 0 255 1000, flo=0, fhi=0.999, dest_type=0 bounding unknown intensity as < 10.3 or > 1461.6 total sample mean = 111.8 (0 zeros) spacing=8, using 2055 sample points, tol=1.00e-03... resetting wm mean[0]: 143 --> 144 resetting gm mean[0]: 89 --> 89 input volume #1 is the most T1-like using real data threshold=11.0 skull bounding box = (49, 43, 9) --> (207, 213, 219) finding center of left hemi white matter using (102, 100, 114) as brain centroid of Right_Cerebral_White_Matter... mean wm in atlas = 144, using box (83,79,88) --> (121, 120,139) to find MRI wm before smoothing, mri peak at 77 robust fit to distribution - 78 +- 5.8 distribution too broad for accurate scaling - disabling WARNING2: gca.c::GCAhistoScaleImageIntensities: h_mri->nbins=113, mri_peak=144 after smoothing, mri peak at 0, scaling input intensities by inf GCAhistoScaleImageIntensities: could not find wm peak Numerical result out of range
I've checked the binary but it doesn't look like this binary contains any reference to FS_HISTO_STD_THRESH parameter.
hayashis@xps15-9560:~/Downloads $ strings mri_deface | grep FS_HISTO hayashis@xps15-9560:~/Downloads 1
Did I download the right binary?
Soichi
On Mon, Aug 5, 2019 at 3:04 PM Bruce Fischl fischl@nmr.mgh.harvard.edu wrote:
no, the env variable won't change that. You can try changing the radius with:
-radius <n>
where n=7 by default. Smaller will make it more aggressive.
cheers Bruce On Mon, 5 Aug 2019, Brian Biekman wrote:
External Email - Use CautionI am also using pediatric brains and am also interesting in using your
defacing algorithm. I downloaded your updated script and changed the environment variable but it leaves a lot of the jaw and chin. Is this something that can be fixed by changing the environment variable to some custom value?
-- Brian Biekman Graduate Student, University of Houston Clinical Psychology, Neuropsychology Concentration Laboratory of Early Experience and Development (LEED) bdbiekman@uh.edu Brian.Biekman@times.uh.edu
From: freesurfer-bounces@nmr.mgh.harvard.edu [
freesurfer-bounces@nmr.mgh.harvard.edu] on behalf of Hoopes, Andrew [ AHOOPES@mgh.harvard.edu]
Sent: Monday, August 05, 2019 12:25 PM To: Freesurfer support list Subject: Re: [Freesurfer] [External] Re: mri_deface fails on pediatric
brain
Hi Soichi, you can update mri_deface with:
curl
https://surfer.nmr.mgh.harvard.edu/pub/dist/freesurfer/dev_binaries/centos7_... -o $FREESURFER_HOME/bin/mri_deface
chmod +x $FREESURFER_HOME/bin/mri_deface
Then, before running mri_deface, set the env variable:
export FS_HISTO_STD_THRESH=20
Then, it should print out something like:
mean wm in atlas = 144, using box (79,77,96) --> (120, 132,144) to find
MRI wm before smoothing, mri peak at 76
FS_HISTO_STD_THRESH found in the environment resetting from 10 to 20.0 robust fit to distribution - 76 +- 6.5 after smoothing, mri peak at 76, scaling input intensities by 1.895 mean wm in atlas = 144, using box (79,77,96) --> (120, 132,144) to find
MRI wm before smoothing, mri peak at 76
FS_HISTO_STD_THRESH found in the environment resetting from 10 to 20.0 robust fit to distribution - 76 +- 6.5 after smoothing, mri peak at 76, scaling input intensities by 1.895
From: freesurfer-bounces@nmr.mgh.harvard.edu <
freesurfer-bounces@nmr.mgh.harvard.edu> on behalf of Soichi Hayashi < hayashis@iu.edu>
Sent: Monday, August 5, 2019 11:32 AM To: Freesurfer support list Subject: Re: [Freesurfer] [External] Re: mri_deface fails on pediatric
brain
External Email - Use CautionWe run on our freesurfer through singularity/docker container, and we
are currently using Centos7 as the base image to build our freesurfer container, but we can run it on any other Linux based OSes. As far as our hardware goes, we run on variety of hardware including HPC system, public and private cloud platforms and a few dedicated desktop machines at our labs.
Thanks! Soichi
On Sun, Aug 4, 2019 at 10:55 PM Bruce Fischl <fischl@nmr.mgh.harvard.edu
mailto:fischl@nmr.mgh.harvard.edu> wrote:
Hi Soichi
I have a fix for you. Can you let us know what kind of hardware/software env you are running in and we will send it to you?
cheers Bruce
On Sun, 4 Aug 2019, Soichi Hayashi wrote:
External Email - Use CautionBruce, I've uploaded one of the failed subject to surfer.nmr.mgh.harvard.edu<
http://surfer.nmr.mgh.harvard.edu%3E ftp
server (named t1.nii.gz).Thank you for looking into this problem.
Soichi
On Sun, Aug 4, 2019 at 10:35 AM Bruce Fischl <
fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu>
wrote: This message was sent from a non-IU address. Please exercise caution when clicking links or opening attachments from external sources. -------
Hi Soichi mri_deface isn't really supported anymore, but if you upload a subject that fails I'll take a look. I would have thought in the ABCD subject age range it would work cheers Bruce On Sat, 3 Aug 2019, Soichi Hayashi wrote: > > External Email - Use Caution > > Hello. > > We are running mri_deface on subjects from ABCD (Adolescent Brain Cognitive > Development) study and many of the subjects are failing on mri_deface. > > $ mri_deface t1.nii.gz > $FREESURFER_HOME/average/talairach_mixed_with_skull.gca > $FREESURFER_HOME/average/face.gca defaced.nii.gz > logging results to defaced.nii.log > reading > '/usr/local/freesurfer/average/talairach_mixed_with_skull.gca'... > reading 't1.nii.gz'... > changing type of input volume to 8 bits/voxel... > MRIchangeType: Building histogram > bounding unknown intensity as < 10.3 or > 1461.6 > total sample mean = 111.8 (0 zeros) > spacing=8, using 2055 sample points, tol=1.00e-03... > resetting wm mean[0]: 143 --> 144 > resetting gm mean[0]: 89 --> 89 > input volume #1 is the most T1-like > using real data threshold=11.0 > skull bounding box = (54, 38, 18) --> (205, 255, 220) > using (104, 110, 119) as brain centroid... > mean wm in atlas = 144, using box (85,83,94) --> (122, 136,143) > to find MRI wm > before smoothing, mri peak at 79 > robust fit to distribution - 80 +- 8.7 > distribution too broad for accurate scaling - disabling > WARNING2: gca.c::GCAhistoScaleImageIntensities: > h_mri->nbins=111, mri_peak=140 > after smoothing, mri peak at 0, scaling input intensities by > inf > GCAhistoScaleImageIntensities: could not find wm peak > Numerical result out of range > > > Should talairach_mixed_with_skull.gca template work with pediatric > subjects? Or is there a different template we must choose? Or maybe any > parameters that we could tweak to get it work better? > > Thanks! > Soichi > >
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Soichi Hayashi