1. It should be very sensitive to this - just make sure the white matter is at least 100 or so in the image (it can be more).
2. You shouldn't need to erase anything - just rerun recon-all
cheers, Bruce On Tue, 29 Apr 2008, Jerry Yeou-Wei Chen wrote:
Okay, thanks! I'll give that a shot... but I have a couple of questions beforehand:
- How would you recommend that I decide what scale factor to use? Trial and
error to get a peak of ~116? 2) Should I run cp and mri_convert, then erase all other existing files and folders, and rerun recon-all from the very beginning (using the new 001.mgz as the input)? Or, can I run cp and mri_convert within the existing subject/mri/orig/ folder and then rerun recon-all (on the same existing file/folders)? I hope my question is clear...!
Thanks,
- Jerry
On Mon, Apr 28, 2008 at 1:52 PM, Bruce Fischl fischl@nmr.mgh.harvard.edu wrote:
Hi Jerry,
I think you can do something like:
cp 001.mgz 001_orig.mgz mri_convert --scale 3 001_orig.mgz 001.mgz
then the same for the other runs and see how that works in recon-all. Usually we detect and account for this, but I guess something is failing in your case.
cheers, Bruce
On Mon, 28 Apr 2008, Jerry Yeou-Wei Chen wrote:
Hi Bruce,
Looking at one of the successfully processed subjects in tkmedit, I would guesstimate the range in the brain to be ~20-65. The corresponding 001.mgz is equally dark.
Regarding the scale option, would I run that initially ( with --scale <scale factor> ), and then run autorecon as normal? Do I find the scale factor as follows: 116 = old peak value * scale factor, where old peak value is the mri peak value attained from these dark images?
Thanks,
- Jerry
On Sun, Apr 27, 2008 at 7:40 AM, Bruce Fischl < fischl@nmr.mgh.harvard.edu> wrote:
p.s. are the 001.mgz and 002.mgz dark? The orig.mgz? Sometimes
nu_correct can create very dark images, and there are some switches you can loo through to correct them. Alternatively if it is the 001.mgz, mri_convert has a scale option you can use.
cheers, Bruce
On Fri, 25 Apr 2008, Jerry Yeou-Wei Chen wrote:
Okay, thanks for your help again Nick!
Regarding the scan data: I am actually working with data collected by a previous member of our lab, and there were indeed some problems... (related to the SCIC algorithm on our GE scanner). So, our technician created a program to "uncorrect" the intensity scaling. The images he saved appear dark when we load them freesurfer, but we expected autorecon's intensity normalization would handle this. In fact, autorecon did successfully process 34 out of 35 scans. (the cortical surface appears quite decent for them, and the thickness values on our average subject appears to be within normal range...) I wonder if I should be concerned about our results due to this intensity issue...
On Tue, Apr 22, 2008 at 2:42 PM, Nick Schmansky < nicks@nmr.mgh.harvard.edu> wrote:
Jerry,
At this point, I think Bruce will need to jump-in to answer (he is out- of-town today). I think he will ask about your scan parameters, ie, how is your scanner setup? It sounds as if something is not quite right in the data coming-off the scanner.
Nick
On Tue, 2008-04-22 at 14:26 -0400, Jerry Yeou-Wei Chen wrote:
Thanks for the reply, Nick. I think you correctly identified the > problem. > nu.mgz is oriented correctly, but looks very dark... However, it > looks > decent if I adjust the brightness/contrast.... > I have other scans that were processed without error, and the > orig.mgz, and nu.mgz also appear quite dark (and the mri peaks > are > very low, too: 29, 46, 52, ...) Also, for those successfully > processed, the brainmask.mgz looks decent as well. > Perhaps this scan that exited with error was too dark to be > corrected > by the intensity normalization algorithm? > > - Jerry > > On Tue, Apr 22, 2008 at 1:22 PM, Nick Schmansky > nicks@nmr.mgh.harvard.edu wrote: > Jerry, > > I'm not sure why it exited without giving a better > message, > but this is > a clue why it exited: > > > before smoothing, mri peak at 0 > after smoothing, mri peak at 0, scaling input intensities > by > inf > > > > > the mri peak is at zero, which is wrong (it should be > around > 116/117 or > so). > > if you view nu.mgz (tkmedit -f nu.mgz), does it look ok? > is > it oriented > correctly? > > Nick > > > > > > On Tue, 2008-04-22 at 12:28 -0400, Jerry Yeou-Wei Chen > wrote: > > Hello, > > > > I am having trouble deciphering why I got the error > attached > below. > > Any suggestions would be greatly appreciated! > > > > Thanks, > > - Jerry > > > > > > > > > > Tue Apr 22 10:15:22 EDT 2008 > > talairach_avi done > > > > cp transforms/talairach.auto.xfm transforms/talairach.xfm > > > > #-------------------------------------------- > > #@# Talairach Failure Detection Tue Apr 22 10:15:23 EDT > 2008 > > /scratch/freesurfer/subjects/17/s017/mri > > > > talairach_afd -T 0.005 -xfm transforms/talairach.xfm > > > > talairach_afd: Talairach Transform: > transforms/talairach.xfm > OK > > (p=0.7072, pval=0.4932 >= threshold=0.0050) > > > > awk - > > > f /scratch/freesurfer/bin/extract_talairach_avi_QA.awk > > /scratch/freesurfer/subjects/17/s017/mri/transforms/talairach_avi.log
>> > TalAviQA: 0.96840 > > z-score: -2 > > #-------------------------------------------- > > #@# Intensity Normalization Tue Apr 22 10:15:23 EDT 2008 > > /scratch/freesurfer/subjects/17/s017/mri > > > > mri_normalize -g 1 nu.mgz T1.mgz > > > > using max gradient = 1.000 > > reading from nu.mgz... > > normalizing image... > > talairach transform > > 0.976 0.083 0.030 -136.866; > > -0.072 1.053 0.164 -153.262; > > -0.020 -0.109 1.085 -122.101; > > 0.000 0.000 0.000 1.000; > > INFO: Modifying talairach volume c_(r,a,s) based on > average_305 > > building Voronoi diagram... > > performing soap bubble smoothing... > > 3d normalization pass 1 of 2 > > building Voronoi diagram... > > performing soap bubble smoothing... > > 3d normalization pass 2 of 2 > > building Voronoi diagram... > > performing soap bubble smoothing... > > writing output to T1.mgz > > 3D bias adjustment took 19 minutes and 44 seconds. > > white matter peak found at 111 > > gm peak at 88 (88), valley at 80 (80) > > csf peak at 57, setting threshold to 77 > > white matter peak found at 111 > > gm peak at 88 (88), valley at 80 (80) > > csf peak at 55, setting threshold to 77 > > #-------------------------------------------- > > #@# Skull Stripping Tue Apr 22 10:35:16 EDT 2008 > > /scratch/freesurfer/subjects/17/s017/mri > > > > mri_em_register -skull > > > nu.mgz > > /scratch/freesurfer/average/RB_all_withskull_2008-03-26.gca
> transforms/talairach_with_skull.lta> > > > aligning to atlas containing skull, setting > unknown_nbr_spacing = 5 > > reading 1 input volumes... > > logging results to talairach_with_skull.log > > reading > > > > > '/scratch/freesurfer/average/RB_all_withskull_2008-03-26.gca'... > > average std = 23.1 using min determinant for > regularization = 53.4 > > 0 singular and 5702 ill-conditioned covariance matrices > regularized > > reading 'nu.mgz'... > > freeing gibbs priors...done. > > bounding unknown intensity as < 20.2 or > 943.7 > > total sample mean = 92.0 (1443 zeros) > > ************************************************ > > spacing=8, using 3481 sample points, tol=1.00e-05... > > ************************************************ > > register_mri: find_optimal_transform > > find_optimal_transform: nsamples 3481, passno 0, spacing 8 > > GCAhistoScaleImageIntensities: could not find wm peak > > resetting wm mean[0]: 117 --> 126 > > resetting gm mean[0]: 74 --> 74 > > input volume #1 is the most T1-like > > using real data threshold=9.0 > > using (102, 110, 122) as brain centroid... > > mean wm in atlas = 126, using box (80,83,98) --> (123, > 136,146) to > > find MRI wm > > before smoothing, mri peak at 0 > > after smoothing, mri peak at 0, scaling input intensities > by > inf > > Linux kdavis8-efef.uhnres.utoronto.ca 2.6.23.1-21.fc7 #1 > SMP > Thu Nov 1 > > 21:09:24 EDT 2007 i686 i686 i386 GNU/Linux > > > > recon-all exited with ERRORS at Tue Apr 22 10:41:40 EDT > 2008 > > > _______________________________________________ > > 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 > >