Hmmm. I looked at your preborder.sh script, and as you said, it includes those key lines.
And here is what happens when I apply them to your borderproj: it.wustl.edu 6% sed 's/Medial/Med/g' Human.003.L.LANDMARKS.updated.borderproj > Human.003.L.LANDMARKS.updated.borderproj.rev it.wustl.edu 7% diff Human.003.L.LANDMARKS.updated.borderproj.rev Human.003.L.LANDMARKS.updated.borderproj 195c195 < 3 117 FLATTEN.HOLE.MedWall 25.000000 1.000000 0.000000 0.000000 --- > 3 117 FLATTEN.HOLE.MedialWall 25.000000 1.000000 0.000000 0.000000 793c793 < 10 62 LANDMARK.MedWall.DORSAL 25.000000 1.000000 0.000000 0.000000 --- > 10 62 LANDMARK.MedialWall.DORSAL 25.000000 1.000000 0.000000 0.000000 857c857 < 11 39 LANDMARK.MedWall.VENTRAL 25.000000 1.000000 0.000000 0.000000 --- > 11 39 LANDMARK.MedialWall.VENTRAL 25.000000 1.000000 0.000000 0.000000 It should have worked. I would like to understand why it did not, which I think will entail your re-running preborder.sh on JUST case 003 and capturing the output for me. But I think it is time to take this thread off-list, because I think this problem is probably rare. Once we figure it out, and I can post the upshot to the list. I will email you offlist about how to capture the output. On 08/11/2010 04:56 AM, Kuk Wai Liu wrote: > Hi Donna, > > I've uploaded my preborder script onto the link you provided me. > > I do have the lines in my preborder script: > > # Sync source border names with target > sed 's/Medial/Med/g' $BORDERPROJ_IN > $BORDERPROJ_IN.rev > mv $BORDERPROJ_IN.rev $BORDERPROJ_IN > > Regards > > Kwliu > ________________________________________ > From: [email protected] > [[email protected]] on behalf of Donna Dierker > [[email protected]] > Sent: 10 August 2010 20:25 > To: Caret, SureFit, and SuMS software users > Subject: Re: [caret-users] Postborder.sh problem > > Kwliu, > > I figured out what is causing the problem. Your borderproj has these > landmarks: > > 10 62 LANDMARK.MedialWall.DORSAL 25.000000 1.000000 0.000000 0.000000 > 11 39 LANDMARK.MedialWall.VENTRAL 25.000000 1.000000 0.000000 0.000000 > > But the target borders are named LANDMARK.MedWall.DORSAL and > LANDMARK.MedWall.VENTRAL. There is a step in the preborder.sh script, > in the loop that does the border variability, that fixes this inconsistency: > > # Sync source border names with target > sed 's/Medial/Med/g' $BORDERPROJ_IN > $BORDERPROJ_IN.rev > mv $BORDERPROJ_IN.rev $BORDERPROJ_IN > > Does your preborder.sh script have those lines in it? I'm wondering if > you are using an older version of preborder.sh, or if you somehow had > the border variability loop disabled. > > Dpmma > > On 08/10/2010 11:00 AM, Donna Dierker wrote: > >> Hmmm. Then I have a hunch what might have happened. Please upload >> your postborder.sh script here: >> >> http://pulvinar.wustl.edu/cgi-bin/upload.cgi >> >> Do *not* email it. >> >> On 08/10/2010 10:22 AM, Kuk Wai Liu wrote: >> >>> Hi Donna, >>> >>> Sorry i was mistaken about there not being a LANDMARK.MedWall.VENTRAL >>> that was missing from the Human.003.L.LANDMARKS.updated.borderproj. >>> I've just taken another look inside the >>> Human.003.L.LANDMARKS.updated.borderproj file and there is a >>> LANDMARK.MedWall.VENTRAL border. >>> >>> I've attached the file with this email. >>> >>> I apologise for my earlier email and for the time you spent analysing >>> it. >>> >>> Regards >>> >>> Kwliu >>> ________________________________________ >>> From: Kuk Wai Liu >>> Sent: 10 August 2010 16:15 >>> To: Caret, SureFit, and SuMS software users >>> Subject: RE: [caret-users] Postborder.sh problem >>> >>> ________________________________________ >>> From: [email protected] >>> [[email protected]] on behalf of Donna Dierker >>> [[email protected]] >>> Sent: 10 August 2010 15:20 >>> To: Caret, SureFit, and SuMS software users >>> Subject: Re: [caret-users] Postborder.sh problem >>> >>> So, we know that Human.003.L.LANDMARKS.borderproj did have a medial wall >>> ventral border, but Human.003.L.LANDMARKS.updated.borderproj, for some >>> reason, did not. (The difference in the filenames is the string >>> "updated." preceding borderproj.) >>> >>> Here is a relevant excerpt from the README.txt: >>> >>> >>> >>>> If any borders do need correction, use Caret's Layers: Border: Border >>>> Update. When the borders are good, use File: Save Data File: >>>> borderproj and >>>> overwrite the borderproj named like *.updated.borderproj. (The >>>> auto-landmarks routine copies Human.SAIS_001.L.LANDMARKS.borderproj to >>>> Human.SAIS_001.L.LANDMARKS.updated.borderproj, so until the borders >>>> are updated, these files are identical. The postborder.sh script uses >>>> the *.updated.borderproj files for registration.) >>>> >>>> >>> This suggests that somehow the medial wall ventral border got deleted >>> during the manual border update process. >>> >>> You can cd 003/surf and launch Caret, opening the left hem spec file and >>> loading scenes. >>> >>> Redo the tweaks, taking care not to delete the medial wall ventral >>> border. Save as Human.003.L.LANDMARKS.updated.borderproj. >>> >>> Re-run postborder.sh on just that subject (i.e., edit subject list to >>> include just 003). >>> >>> I can't think of a more straightforward way of doing it. >>> >>> If you spent a ton of time tweaking the borders in >>> Human.003.L.LANDMARKS.updated.borderproj, and don't want to have to >>> repeat those edits, then there are ways of salvaging them. But I'm >>> guessing that is not likely. >>> >>> On 08/10/2010 06:29 AM, Kuk Wai Liu wrote: >>> >>> >>>> Hi Donna, >>>> >>>> It appears that you are correct about the LANDMARK.MedWall.VENTRAL >>>> that is missing from the Human.003.L.LANDMARKS.updated.borderproj. >>>> What would i need to do to generate one? >>>> >>>> There does appear to be a medial wall ventral border capture in the >>>> html directory. >>>> >>>> I've attached the images i've obtained for one subject (including >>>> both left and right hemispheres) after running the preborder script. >>>> >>>> Regards >>>> >>>> Kwliu >>>> ________________________________________ >>>> From: [email protected] >>>> [[email protected]] on behalf of Donna Dierker >>>> [[email protected]] >>>> Sent: 09 August 2010 16:48 >>>> To: Caret, SureFit, and SuMS software users >>>> Subject: Re: [caret-users] Postborder.sh problem >>>> >>>> Kwliu, >>>> >>>> It appears that Human.003.L.LANDMARKS.updated.borderproj is missing >>>> LANDMARK.MedWall.VENTRAL, which causes all the downstream trouble. >>>> >>>> It is possible that the auto-landmark algorithm had trouble >>>> identifying this landmark (although in our experience this is >>>> rare). Or it is possible that it got deleted during tweaking. >>>> >>>> Or redrawn using the border name ATLAS_LANDMARK.MedWall.VENTRAL >>>> instead of just LANDMARK.MedWall.VENTRAL. (All borders named like >>>> "ATLAS_*" get deleted by the postborder.sh script. They are useful >>>> for showing where the target borders lie, though.) >>>> >>>> Did the captures generated by preborder.sh (in the HTMLDIR) show a >>>> medial wall ventral border? >>>> >>>> Donna >>>> >>>> >>>> On 08/09/2010 06:18 AM, Kuk Wai Liu wrote: >>>> >>>> >>>> >>>>> Hi, >>>>> >>>>> Thanks for your help on the preborder.sh problem. Having run the >>>>> preborder.sh command successfully without errors, i've now got errors >>>>> in the postborder.sh command, could you help me out please? My output >>>>> is attached with this email. >>>>> >>>>> Regards >>>>> >>>>> Kwliu >>>>> > > _______________________________________________ > caret-users mailing list > [email protected] > http://brainvis.wustl.edu/mailman/listinfo/caret-users > > _______________________________________________ > caret-users mailing list > [email protected] > http://brainvis.wustl.edu/mailman/listinfo/caret-users > _______________________________________________ caret-users mailing list [email protected] http://brainvis.wustl.edu/mailman/listinfo/caret-users
