Try installing the attached file. Cliff MacNab previously reported a
similar problem, and the fix is waiting to be released among the next
set of spm8 updates.
Best regards,
-John
On Wed, 2009-07-15 at 14:55 -0500, Pierson, Ronald K wrote:
> I have had no problems since the initial attempt; I have not been able
> to replicate my problem. So, my only guess is that it was an operator
> error - maybe I replicated the tissue classes by mistake, in addition to
> the channels. Sorry for the false alarm.
>
> I am running over 1000 scans through, all with T1 and T2 images that are
> 1 mm cubic voxels, 256x256x256. For a few of the scans I get the
> following error, about 3 iterations into the process (based on the
> graphic). These scans will not get past this point, even after
> retrying.
>
> Any ideas on what the error refers to, or what I should try to get them
> running?
>
> ########
>
>
> Running 'New Segment'
> Failed 'New Segment'
> Error using ==> times
> Matrix dimensions must agree.
> In file "/IPLlinux/raid0/homes/ronald/SPM8/spm8/toolbox/Seg/spm_maff8.m"
> (v2281), function "affreg" at line 157.
> In file "/IPLlinux/raid0/homes/ronald/SPM8/spm8/toolbox/Seg/spm_maff8.m"
> (v2281), function "spm_maff8" at line 27.
> In file
> "/IPLlinux/raid0/homes/ronald/SPM8/spm8/toolbox/Seg/spm_preproc_run.m"
> (v2281), function "run_job" at line 80.
> In file
> "/IPLlinux/raid0/homes/ronald/SPM8/spm8/toolbox/Seg/spm_preproc_run.m"
> (v2281), function "spm_preproc_run" at line 27.
>
> The following modules did not run:
> Failed: New Segment
>
> ############
>
> Ron
>
>
> -----Original Message-----
> From: John Ashburner [mailto:[log in to unmask]]
> Sent: Friday, July 10, 2009 5:42 AM
> To: Pierson, Ronald K
> Cc: [log in to unmask]
> Subject: RE: [SPM] New Segment on T1 and T2, brain-only images
>
> I have no idea what has happened here. Unless the number of tissue
> classes is changed, there should be only six classes (c*.nii) qritten
> out at the most. Could you email me the batch job so I can take a look?
>
> Alternatively, it could be a bug that I have not yet experienced.
>
> Best regards,
> -John
>
> On Thu, 2009-07-09 at 16:12 -0500, Pierson, Ronald K wrote:
> > Here are some snapshots of my T1 T2 images, and my results after going
> > with both 6 tissue classes and 3 tissue classes.
> >
> > When I use the default TPMs (spm8/toolbox/Seg/tpm.nii), which uses 6
> > tissues as far as I understand, I get the 6class images, but 9 of them
> -
> > C1, C2, C3, C4, C5, C7, C8, C9, C10. C1 and C7 both appear to be
> > versions of white matter, but at different ends of the spectrum. C2
> and
> > C8 seem to both be WM, but again neither one is complete on its own.
> C3
> > and C9 are CSF, and the rest are edge voxels.
> >
> > When I use only the first 3 classes I get the 3class images C1, C2,
> and
> > C3, which seem to be pretty good GM, WM and CSF images.
> >
> > So, I guess I am concerned that with the 6 classes I am getting two
> > separate GM images, neither of which is complete on its own, and I
> would
> > think I should only use one. Am I making an assumption that isn't
> > correct? Or does DARTEL know how to handle this correctly, and I
> should
> > just move on...
> >
> > Ron
> >
> >
> >
> > -----Original Message-----
> > From: John Ashburner [mailto:[log in to unmask]]
> > Sent: Thursday, July 09, 2009 3:44 PM
> > To: Pierson, Ronald K
> > Cc: [log in to unmask]
> > Subject: Re: [SPM] New Segment on T1 and T2, brain-only images
> >
> > You should be able to run the new segmentation on the skull-stripped
> > data without any tweeks, providing the stripped off voxels are set to
> > either 0 or NaN. Values of 0 or NaN in the image will be ignored.
> >
> > Best regards,
> > -John
> >
> > On Thu, 2009-07-09 at 18:37 +0100, Ronald Pierson wrote:
> > > I want to preprocess some scans for VBM using new segment and
> DARTEL.
> > I
> > > want to use both T1 and T2 scans, and they are currently
> > brain-clipped. In
> > > new segment can I run it with just the first three tissue classes,
> > since there
> > > are no non-brain tissues present? Any other suggestions, or is
> there
> > a
> > > tutorial/procedure manual available that would be useful for me to
> > look at?
> > >
> > > Ronald Pierson
> > > Chair of Imaging, PREDICT-HD
> > > Manager, Research and Development
> > > Department of Psychiatry, Carver College of Medicine
> > > W-278 GH, 200 Hawkins Dr
> > > Iowa City, IA 52242
> > >
--
John Ashburner <[log in to unmask]>
|