Hi everyone, I'm using recon-all to flatten a number of scans from patients and controls. Most of them work fine, but there are 3 that do not reconstruct correctly. They all seem to share the same problem: Recon-all aborts during the automatic topology correction step, and there's a line in the log saying mrisDefectAnalysis : defect XXX overlap defect YYY. I've checked registration to talairach, skull strip and initial surfaces, the only thing I could see was that in some cases, bits of the superior sagittal sinus were wrongly included in the lh.orig.nofix, or small bits of dura, but nothing major. I'm attaching the relevant output from the topology fixer for two different cases. I would be grateful if someone could point me at other things to try. Is it possible to visualise the defects' locations, so I could find out which defect is actually causing the overlap message? Thanks for your help everyone, Johannes
----------
CASE 1 #@# Fix Topology lh Wed Jun 14 04:04:29 BST 2006 \n cp ../surf/lh.orig.nofix ../surf/lh.orig \n \n cp ../surf/lh.inflated.nofix ../surf/lh.inflated \n /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_2280/scripts \n mris_fix_topology -mgz -sphere qsphere.nofix -ga PAT_2280 lh \n reading spherical homeomorphism from 'qsphere.nofix' using genetic algorithm with optimized parameters
************************************************************* Topology Correction Parameters retessellation mode: genetic search number of patches/generation : 10 number of generations : 10 surface mri loglikelihood coefficient : 1.0 volume mri loglikelihood coefficient : 10.0 normal dot loglikelihood coefficient : 1.0 quadratic curvature loglikelihood coefficient : 1.0 volume resolution : 2 eliminate vertices during search : 1 initial patch selection : 1 select all defect vertices : 0 ordering dependant retessellation: 0 use precomputed edge table : 0 smooth retessellated patch : 2 match retessellated patch : 1 verbose mode : 0
************************************************************* INFO: assuming .mgz format $Id: mris_fix_topology.c,v 1.41 2006/02/06 19:41:59 nicks Exp $ $Id: mrisurf.c,v 1.441.2.1 2006/03/06 21:12:31 greve Exp $ before topology correction, eno=-656 (nv=197397, nf=396220, ne=594273, g=329) using quasi-homeomorphic spherical map to tessellate cortical surface...
Correction of the Topology Finding true center and radius of Spherical Surface...done Surface centered at (0,0,0) with radius 100.0 in 9 iterations marking ambiguous vertices... 30743 ambiguous faces found in tessellation segmenting defects... defect 6 : 7 vertices have been added to the surface
[ ... ]
defect 314 : 2 vertices have been added to the surface total of 95 vertices has been added to the surface analyzing defects... mrisDefectAnalysis : defect 277 overlap defect 272 reading input surface /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_2280/surf/lh.qsphere.nofix... reading brain volume from brain... reading wm segmentation from wm... Darwin jalapeno6.fmrib.ox.ac.uk 8.6.0 Darwin Kernel Version 8.6.0: Tue Mar 7 16:58:48 PST 2006; root:xnu-792.6.70.obj~1/RELEASE_PPC Power Macintosh powerpc
recon-all exited with ERRORS at Wed Jun 14 04:08:10 BST 2006
CASE 2 #@# Fix Topology rh Thu Jun 15 12:13:05 BST 2006 \n cp ../surf/rh.orig.nofix ../surf/rh.orig \n \n cp ../surf/rh.inflated.nofix ../surf/rh.inflated \n /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_3509/scripts \n mris_fix_topology -mgz -sphere qsphere.nofix -ga PAT_3509 rh \n reading spherical homeomorphism from 'qsphere.nofix' using genetic algorithm with optimized parameters
************************************************************* Topology Correction Parameters retessellation mode: genetic search number of patches/generation : 10 number of generations : 10 surface mri loglikelihood coefficient : 1.0 volume mri loglikelihood coefficient : 10.0 normal dot loglikelihood coefficient : 1.0 quadratic curvature loglikelihood coefficient : 1.0 volume resolution : 2 eliminate vertices during search : 1 initial patch selection : 1 select all defect vertices : 0 ordering dependant retessellation: 0 use precomputed edge table : 0 smooth retessellated patch : 2 match retessellated patch : 1 verbose mode : 0
************************************************************* INFO: assuming .mgz format $Id: mris_fix_topology.c,v 1.41 2006/02/06 19:41:59 nicks Exp $ $Id: mrisurf.c,v 1.441.2.1 2006/03/06 21:12:31 greve Exp $ before topology correction, eno=-2025 (nv=252518, nf=509180, ne=763723, g=1013) using quasi-homeomorphic spherical map to tessellate cortical surface...
Correction of the Topology Finding true center and radius of Spherical Surface...done Surface centered at (0,0,0) with radius 100.0 in 13 iterations marking ambiguous vertices... 137391 ambiguous faces found in tessellation segmenting defects... defect 12 : 4 vertices have been added to the surface
[ ... ]
defect 440 : 4 vertices have been added to the surface total of 340 vertices has been added to the surface analyzing defects... mrisDefectAnalysis : defect 395 overlap defect 393 reading input surface /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_3509/surf/rh.qsphere.nofix... reading brain volume from brain... reading wm segmentation from wm... Darwin jalapeno4.fmrib.ox.ac.uk 8.6.0 Darwin Kernel Version 8.6.0: Tue Mar 7 16:58:48 PST 2006; root:xnu-792.6.70.obj~1/RELEASE_PPC Power Macintosh powerpc
recon-all exited with ERRORS at Thu Jun 15 12:19:26 BST 2006
Johannes,
To visualize defect locations:
tksurfer <subjid> lh inflated.nofix
then:
File->Curvature->Load Curvature
select:
surf/lh.defect_labels
and the defects will display in red.
Remember you can click on a defect vertex in tksurfer, then click 'save point' and then in tkmedit, click 'goto saved point', which is handy to see exactly where in a volume things went wrong.
Nick
On Wed, 2006-07-12 at 17:55 +0100, Johannes Klein wrote:
Hi everyone, I'm using recon-all to flatten a number of scans from patients and controls. Most of them work fine, but there are 3 that do not reconstruct correctly. They all seem to share the same problem: Recon-all aborts during the automatic topology correction step, and there's a line in the log saying mrisDefectAnalysis : defect XXX overlap defect YYY. I've checked registration to talairach, skull strip and initial surfaces, the only thing I could see was that in some cases, bits of the superior sagittal sinus were wrongly included in the lh.orig.nofix, or small bits of dura, but nothing major. I'm attaching the relevant output from the topology fixer for two different cases. I would be grateful if someone could point me at other things to try. Is it possible to visualise the defects' locations, so I could find out which defect is actually causing the overlap message? Thanks for your help everyone, Johannes
CASE 1 #@# Fix Topology lh Wed Jun 14 04:04:29 BST 2006 \n cp ../surf/lh.orig.nofix ../surf/lh.orig \n \n cp ../surf/lh.inflated.nofix ../surf/lh.inflated \n /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_2280/scripts \n mris_fix_topology -mgz -sphere qsphere.nofix -ga PAT_2280 lh \n reading spherical homeomorphism from 'qsphere.nofix' using genetic algorithm with optimized parameters
Topology Correction Parameters retessellation mode: genetic search number of patches/generation : 10 number of generations : 10 surface mri loglikelihood coefficient : 1.0 volume mri loglikelihood coefficient : 10.0 normal dot loglikelihood coefficient : 1.0 quadratic curvature loglikelihood coefficient : 1.0 volume resolution : 2 eliminate vertices during search : 1 initial patch selection : 1 select all defect vertices : 0 ordering dependant retessellation: 0 use precomputed edge table : 0 smooth retessellated patch : 2 match retessellated patch : 1 verbose mode : 0
INFO: assuming .mgz format $Id: mris_fix_topology.c,v 1.41 2006/02/06 19:41:59 nicks Exp $ $Id: mrisurf.c,v 1.441.2.1 2006/03/06 21:12:31 greve Exp $ before topology correction, eno=-656 (nv=197397, nf=396220, ne=594273, g=329) using quasi-homeomorphic spherical map to tessellate cortical surface...
Correction of the Topology Finding true center and radius of Spherical Surface...done Surface centered at (0,0,0) with radius 100.0 in 9 iterations marking ambiguous vertices... 30743 ambiguous faces found in tessellation segmenting defects... defect 6 : 7 vertices have been added to the surface
[ ... ]
defect 314 : 2 vertices have been added to the surface total of 95 vertices has been added to the surfaceanalyzing defects... mrisDefectAnalysis : defect 277 overlap defect 272 reading input surface /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_2280/surf/lh.qsphere.nofix... reading brain volume from brain... reading wm segmentation from wm... Darwin jalapeno6.fmrib.ox.ac.uk 8.6.0 Darwin Kernel Version 8.6.0: Tue Mar 7 16:58:48 PST 2006; root:xnu-792.6.70.obj~1/RELEASE_PPC Power Macintosh powerpc
recon-all exited with ERRORS at Wed Jun 14 04:08:10 BST 2006
CASE 2 #@# Fix Topology rh Thu Jun 15 12:13:05 BST 2006 \n cp ../surf/rh.orig.nofix ../surf/rh.orig \n \n cp ../surf/rh.inflated.nofix ../surf/rh.inflated \n /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_3509/scripts \n mris_fix_topology -mgz -sphere qsphere.nofix -ga PAT_3509 rh \n reading spherical homeomorphism from 'qsphere.nofix' using genetic algorithm with optimized parameters
Topology Correction Parameters retessellation mode: genetic search number of patches/generation : 10 number of generations : 10 surface mri loglikelihood coefficient : 1.0 volume mri loglikelihood coefficient : 10.0 normal dot loglikelihood coefficient : 1.0 quadratic curvature loglikelihood coefficient : 1.0 volume resolution : 2 eliminate vertices during search : 1 initial patch selection : 1 select all defect vertices : 0 ordering dependant retessellation: 0 use precomputed edge table : 0 smooth retessellated patch : 2 match retessellated patch : 1 verbose mode : 0
INFO: assuming .mgz format $Id: mris_fix_topology.c,v 1.41 2006/02/06 19:41:59 nicks Exp $ $Id: mrisurf.c,v 1.441.2.1 2006/03/06 21:12:31 greve Exp $ before topology correction, eno=-2025 (nv=252518, nf=509180, ne=763723, g=1013) using quasi-homeomorphic spherical map to tessellate cortical surface...
Correction of the Topology Finding true center and radius of Spherical Surface...done Surface centered at (0,0,0) with radius 100.0 in 13 iterations marking ambiguous vertices... 137391 ambiguous faces found in tessellation segmenting defects... defect 12 : 4 vertices have been added to the surface
[ ... ]
defect 440 : 4 vertices have been added to the surface total of 340 vertices has been added to the surfaceanalyzing defects... mrisDefectAnalysis : defect 395 overlap defect 393 reading input surface /Volumes/JalapenoSAN/Scratch/natalie/schizophrenia/fs/PAT_3509/surf/rh.qsphere.nofix... reading brain volume from brain... reading wm segmentation from wm... Darwin jalapeno4.fmrib.ox.ac.uk 8.6.0 Darwin Kernel Version 8.6.0: Tue Mar 7 16:58:48 PST 2006; root:xnu-792.6.70.obj~1/RELEASE_PPC Power Macintosh powerpc
recon-all exited with ERRORS at Thu Jun 15 12:19:26 BST 2006
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi Nick, Thanks for your message. Unfortunately, the topology fixer did not produce a defect_labels file, it seems to have aborted before doing so. Is there a way to make mris_fix_topology produce the defect labels without actually running topology correction? I thought of trying the -correct_defect option with one of the defects that didn't cause problems. Does that make sense? Thanks, Johannes
Nick Schmansky schrieb:
Johannes,
To visualize defect locations:
tksurfer <subjid> lh inflated.nofix
then:
File->Curvature->Load Curvature
select:
surf/lh.defect_labels
and the defects will display in red.
Remember you can click on a defect vertex in tksurfer, then click 'save point' and then in tkmedit, click 'goto saved point', which is handy to see exactly where in a volume things went wrong.
Nick
Johannes,
In looking at your recon-all.log, it looks like you are running v3.0.1 of freesurfer, which can be confirmed by typing:
cat $FREESURFER_HOME/build-stamp.txt
In v3.0.2, a bug in the topology fixer was fixed (namely, the 'mrisDefectAnalysis' routine).
Is it possible for you to download the latest freesurfer and rerun on the trouble subjects? It should not be necessary to re-run on the other subjects. The fixed topology fixer is likely to successfully process your subjects.
I can also send you this new mris_topology_fixer binary for the Mac if you don't want to download/re-install.
Nick
On Wed, 2006-07-12 at 18:50 +0100, Johannes Klein wrote:
Hi Nick, Thanks for your message. Unfortunately, the topology fixer did not produce a defect_labels file, it seems to have aborted before doing so. Is there a way to make mris_fix_topology produce the defect labels without actually running topology correction? I thought of trying the -correct_defect option with one of the defects that didn't cause problems. Does that make sense? Thanks, Johannes
Nick Schmansky schrieb:
Johannes,
To visualize defect locations:
tksurfer <subjid> lh inflated.nofix
then:
File->Curvature->Load Curvature
select:
surf/lh.defect_labels
and the defects will display in red.
Remember you can click on a defect vertex in tksurfer, then click 'save point' and then in tkmedit, click 'goto saved point', which is handy to see exactly where in a volume things went wrong.
Nick
Hi Nick, hi everyone, I just ran the 3.0.2 topology fixer on a linux machine over one of the failed surfaces, and that seems to have worked - great! The compute servers have since been upgraded to 3.0.3, so I'll feed the results into the rest of the recon-all pipeline and see how it goes. Thanks for all your help, Johannes
Nick Schmansky schrieb:
Johannes,
In looking at your recon-all.log, it looks like you are running v3.0.1 of freesurfer, which can be confirmed by typing:
cat $FREESURFER_HOME/build-stamp.txt
In v3.0.2, a bug in the topology fixer was fixed (namely, the 'mrisDefectAnalysis' routine).
Is it possible for you to download the latest freesurfer and rerun on the trouble subjects? It should not be necessary to re-run on the other subjects. The fixed topology fixer is likely to successfully process your subjects.
I can also send you this new mris_topology_fixer binary for the Mac if you don't want to download/re-install.
Nick
On Wed, 2006-07-12 at 18:50 +0100, Johannes Klein wrote:
Hi Nick, Thanks for your message. Unfortunately, the topology fixer did not produce a defect_labels file, it seems to have aborted before doing so. Is there a way to make mris_fix_topology produce the defect labels without actually running topology correction? I thought of trying the -correct_defect option with one of the defects that didn't cause problems. Does that make sense? Thanks, Johannes
Nick Schmansky schrieb:
Johannes,
To visualize defect locations:
tksurfer <subjid> lh inflated.nofix
then:
File->Curvature->Load Curvature
select:
surf/lh.defect_labels
and the defects will display in red.
Remember you can click on a defect vertex in tksurfer, then click 'save point' and then in tkmedit, click 'goto saved point', which is handy to see exactly where in a volume things went wrong.
Nick
Hi Johannes,
I think you can run it with -diagonly and it will quit after identifying the defects (should just take a couple of minutes). I thought Florent had fixed this problem. What version are you running? We may need to wait for Flo to help with this....
Bruce
On Wed, 12 Jul 2006, Johannes Klein wrote:
Hi Nick, Thanks for your message. Unfortunately, the topology fixer did not produce a defect_labels file, it seems to have aborted before doing so. Is there a way to make mris_fix_topology produce the defect labels without actually running topology correction? I thought of trying the -correct_defect option with one of the defects that didn't cause problems. Does that make sense? Thanks, Johannes
Nick Schmansky schrieb:
Johannes,
To visualize defect locations:
tksurfer <subjid> lh inflated.nofix
then:
File->Curvature->Load Curvature
select:
surf/lh.defect_labels
and the defects will display in red.
Remember you can click on a defect vertex in tksurfer, then click 'save point' and then in tkmedit, click 'goto saved point', which is handy to see exactly where in a volume things went wrong.
Nick
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Johannes Klein -
Nick Schmansky