Hello Mark,
Thanks for the quick reply;
I just saw that the -s option is explained in the First manual on the
website as well; must have overlooked it, sorry; anyway thank you for
the reminder. The script runs without any problems now (without the
brainstem).
Jasper
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf Of Mark Jenkinson
Sent: woensdag 14 oktober 2009 0:00
To: [log in to unmask]
Subject: Re: [FSL] FIRST-brainstem segmentation failure
Hi,
I think you are probably right about the reason for it failing.
There is a cleanup stage following the individual segmentations
which deals with overlaps, so it is most useful to keep using
run_first_all rather than do things separately. However, the
way to get everything without the brainstem is simple. It is:
run_first_all -i input -o output -s
L_Accu
,L_Amyg
,L_Caud
,L_Hipp
,L_Late
,L_Pall
,L_Puta,L_Thal,R_Accu,R_Amyg,R_Caud,R_Hipp,R_Late,R_Pall,R_Puta,R_Thal
(this should all be on one line and with a space after the -s, but no
spaces between the structure names - only commas).
This will then do what you want.
Note that you can do any subset in this way.
All the best,
Mark
On 12 Oct 2009, at 09:04, Jasper Luyendijk wrote:
> Hi,
>
> I have encountered a problem with the segmentation of the brainstem
> using
> FIRST. A few images of my 3mm 3D T1W-image dataset (3T) gave no
> _all_fast_firstseg and _all_fast_origsegs output because the script
> failed at
> the segmentation of the brainstem.
> I've attached the relevant error reports. I've seen the former
> reports on
> brainstem segmentation failure and I think I might have the same
> problem; in
> the registration output image of the missegmented images a small
> part of the
> image has been lost at the corners of the FOV. (I think because in
> the original
> image the top of the skull and the lower part of the cerebellum and
> medulla
> oblongata was left out in the acquisition). The rest of the
> _to_std_sub image
> is completely nrmal. Due to this registration complication the lower
> part of the
> brainstem has been cut off in the failed subjects. I think that that
> is the
> reason why the segmentation failed (application of the brainstem
> model not
> possible). In the subjects with succesfully segmented brainstems, the
> brainstem was not affected by this registration complication.
> (It could also be that the original T1W scans have to little
> brainstem to begin
> with). It appears from the output.logs that all the other structures
> were
> segmented without problems (no error reports and the registration of
> the basal
> ganglia is just fine.), but since the scripts deletes all the separate
> segmentation output images (structure_corr) and needs the brainstem to
> create the final 4D output, I don't have them.
> My question is; Since I'm not interested in the brainstem but only
> in the basal
> ganglia; couldn't I just perform the separate first commado's that
> are given by
> the output.com, -2,-3 files after the unsuccesful run, and leave the
> brainstem
> out? I prefer this solution because the program would select the
> best modes of
> segmentation for me. Or does the script do something with the output
> that is
> not reflected in the output.com files? Or is the segmentation of the
> other
> structures somehow affected by the unsuccesful segmentation of the
> brainstem?
> If you know even simpeler solutions I'm very much interested as well.
>
> Thanks, kind regards,
>
> Jasper Luyendijk
>
> Leiden University Medical Center
>
>
> /opt/fsl-4.1.4/bin/run_first: line 160: 12113 Segmentation
> fault ${FSLDIR}/bin/first -i $image -l $trmat -m $model -k
> $outname -n $modes $verbose $comExtras
>
>
> ** ERROR (nifti_image_read): failed to find header file for
> 'image_output-BrStem_corr'
> ** ERROR: nifti_image_open(image_output-BrStem_corr): bad header info
> Error: failed to open file image_output-BrStem_corr
> ERROR: Could not open image image_output-BrStem_corr
> Image Exception : #22 :: Failed to read volume image_output-
> BrStem_corr
> terminate called after throwing an instance of
> 'RBD_COMMON::BaseException'
> /bin/sh: line 1: 12157 Aborted /opt/fsl-4.1.4/bin/
> fslmerge -t image_output_all_fast_firstseg image_output-L_Accu_corr
> image_output-L_Amyg_corr image_output-L_Caud_corr image_output-
> L_Hipp_corr image_output-L_Late_corr image_output-L_Pall_corr
> image_output-L_Puta_corr image_output-L_Thal_corr image_output-
> R_Accu_corr image_output-R_Amyg_corr image_output-R_Caud_corr
> image_output-R_Hipp_corr image_output-R_Late_corr image_output-
> R_Pall_corr image_output-R_Puta_corr image_output-R_Thal_corr
> image_output-BrStem_corr
> ** ERROR (nifti_image_read): failed to find header file for
> 'image_output-BrStem_first'
> ** ERROR: nifti_image_open(image_output-BrStem_first): bad header info
> Error: failed to open file image_output-BrStem_first
> ERROR: Could not open image image_output-BrStem_first
> Image Exception : #22 :: Failed to read volume image_output-
> BrStem_first
> terminate called after throwing an instance of
> 'RBD_COMMON::BaseException'
> /bin/sh: line 1: 12158 Aborted /opt/fsl-4.1.4/bin/
> fslmerge -t image_output_all_fast_origsegs image_output-L_Accu_first
> image_output-L_Amyg_first image_output-L_Caud_first image_output-
> L_Hipp_first image_output-L_Late_first image_output-L_Pall_first
> image_output-L_Puta_first image_output-L_Thal_first image_output-
> R_Accu_first image_output-R_Amyg_first image_output-R_Caud_first
> image_output-R_Hipp_first image_output-R_Late_first image_output-
> R_Pall_first image_output-R_Puta_first image_output-R_Thal_first
> image_output-BrStem_first
> ** ERROR (nifti_image_read): failed to find header file for
> 'image_output_all_fast_origsegs'
> ** ERROR: nifti_image_open(image_output_all_fast_origsegs): bad
> header info
> Error: failed to open file image_output_all_fast_origsegs
> ERROR: Could not open image image_output_all_fast_origsegs
> Image Exception : #22 :: Failed to read volume
> image_output_all_fast_origsegs
> Aborted
>
>
|