Colin, You might consider the following.
Draw borders around the medial wall (as you've already done), and paint as many nodes as you can as 'medial-wall'. Select Surface ROI: Paint: medial-wall, then select ROI operation: disconnect selected nodes. Go back to Surface ROI and select Remove Islands. THis should remove ALL of the medial wall, unless there is a 'finger' that connects some interior medial wall to the rest of cortex. If there is such a finger, use Surface: Cuts: Draw cuts to make a full disconnect, then repeat the remove islands step. This should get you a surface in which ALL medial-wall nodes are disconnected. Save the topology file as ***.open.topo Then select all (non-medial-wall) nodes and give them all some interim assignment. (e.g. 'temp-cortex') in a new paint column. The remaining unassigned nodes (all '???') should constitute a 'true' medial-wall.'' Save this paint file approrpiately. Return to a surface in which you have the original closed topo file. Using Surface ROI, select the '???' nodes from the above paint file. If they constitute all of the true medial-wall nodes, then you should be able to smooth them all and achieve your objective. Believe me, I recognize that the devil's in the details. I'm not certain that this will work, but I suspect that it (or some variation therof) will finally get the job done for you. David On Jul 21, 2012, at 3:18 PM, Colin Reveley wrote: > the only thing that works without making a new surface is smoothing > operations on parts of the ROI, strategically, till smoothing on the wole > thing might work > > but this is also tricksy. it's generally easier to to do by hand altering > the segmentation to remove the biggest issues. this applies to surfaces made > in caret with surefit as well as FS stuff, although less so. > > On 21 July 2012 21:09, Colin Reveley <[email protected]> wrote: > in may cases it is very, VERY hard to paint the apropriate subset of nodes, > if the geometry is truly complex. > > one must use many little borders instead of one large one. > > even if you do get all the nodes painted, smoothing operations won't work out > in these situtations. > > this isn;t a niave question, to which the standard solutions apply. I've > spent probably hundreds of hours on this issue. it is very hard. It may well > only come in in ex-vivo. > > caret has problems with painting by border enclusure, and with smoothing, > when the geometry of the set of nodes is really complex, convex, concave, and > folder UNDER itself (but still homeomophic to a sphere) > > I thought you guys probably knew that. > > On 21 July 2012 18:00, <[email protected]> wrote: > Send caret-users mailing list submissions to > [email protected] > > To subscribe or unsubscribe via the World Wide Web, visit > http://brainvis.wustl.edu/mailman/listinfo/caret-users > or, via email, send a message with subject or body 'help' to > [email protected] > > You can reach the person managing the list at > [email protected] > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of caret-users digest..." > > Today's Topics: > > 1. Re: part of a surface (Donna Dierker) > 2. Re: part of a surface (David Van Essen) > 3. Re: Freesurfer Surface and Volume are not in register when > shown in Caret (Pablo Polosecki) > 4. Re: Freesurfer Surface and Volume are not in register when > shown in Caret (David Van Essen) > > > ---------- Forwarded message ---------- > From: Donna Dierker <[email protected]> > To: "Caret, SureFit, and SuMS software users" <[email protected]> > Cc: > Date: Fri, 20 Jul 2012 18:18:51 -0500 > Subject: Re: [caret-users] part of a surface > On Jul 20, 2012, at 5:06 PM, Colin Reveley wrote: > > > Is it possible to inflate just part of a surface? I doubt it. > > Inflate, I don't think so; smooth -- definitely. You can use paint to select > nodes, then constrain the smoothing that way. > > I don't have caret handy, but look at the options under Surface: Geometry. > > > > > I am experiencing my usual agony in smoothing the medial wall of my > > surfaces. as you can see, the medial wall region is very complicated and > > hard to make into a smooth sheet, with big convexities, big concavities, > > and loads of smaller ones. > > > > Actually I can't even make a crossover free sphere from this one. > > > > yes, I can change my segmentation by hand. but at the risk of losing > > accuracy. altering the segmentation is normally what I do. aside from being > > a bit dodgy, it's also very laborious. > > > > another way is to smooth it piecemeal, but this example is probably too far > > gone for that. > > > > can't I just suck those nodes out of there?! > > > > best, > > > > Colin > > <capture.jpg>_______________________________________________ > > caret-users mailing list > > [email protected] > > http://brainvis.wustl.edu/mailman/listinfo/caret-users > > > > > > ---------- Forwarded message ---------- > From: David Van Essen <[email protected]> > To: "Caret, SureFit, and SuMS software users" <[email protected]> > Cc: > Date: Fri, 20 Jul 2012 18:20:54 -0500 > Subject: Re: [caret-users] part of a surface > Colin, > > First, assign your medial wall nodes to a paint (label) identity. If you > haven't done this already, it can be done using Draw Borders and filling in > the ROI, starting with a very inflated or spherical surface. If you have > crossovers in your medial wall, this might take some futzing and an extra > round, but I think it can be made to work. > > Then use Surface: Surface ROI: Node Selection: Paint, and select the medial > wall nodes, but starting with the midthickness configuration in the main > window. > Then use ROI Operation: Smoothing, and choose parameters that yield adequate > smoothing. > > If you have crossovers initially, you can probably re-inflate and re-project > to the sphere starting with the partially smoothed medial wall. > > Good luck. > > David > > > On Jul 20, 2012, at 5:06 PM, Colin Reveley wrote: > > > Is it possible to inflate just part of a surface? I doubt it. > > > > I am experiencing my usual agony in smoothing the medial wall of my > > surfaces. as you can see, the medial wall region is very complicated and > > hard to make into a smooth sheet, with big convexities, big concavities, > > and loads of smaller ones. > > > > Actually I can't even make a crossover free sphere from this one. > > > > yes, I can change my segmentation by hand. but at the risk of losing > > accuracy. altering the segmentation is normally what I do. aside from being > > a bit dodgy, it's also very laborious. > > > > another way is to smooth it piecemeal, but this example is probably too far > > gone for that. > > > > can't I just suck those nodes out of there?! > > > > best, > > > > Colin > > <capture.jpg>_______________________________________________ > > caret-users mailing list > > [email protected] > > http://brainvis.wustl.edu/mailman/listinfo/caret-users > > > > > > ---------- Forwarded message ---------- > From: Pablo Polosecki <[email protected]> > To: "Caret, SureFit, and SuMS software users" <[email protected]> > Cc: > Date: Fri, 20 Jul 2012 19:27:48 -0400 > Subject: Re: [caret-users] Freesurfer Surface and Volume are not in register > when shown in Caret > Hi Donna, > > Your excerpts were really inspiring. Many many thanks. I got them > finally well-alligned. > One more question: My volumes and surfaces have a nominal resolution > of 1mm but they are actually .5mm (This means we scan at .5 mm > resolution but we have to trick FreeSurfer into thinking we scan at > 1mm because it doesn't display things correctly with other resolution > values). Does this matter for working in Caret and registering to F99? > Or do I have to rescale all my files back to .5mm? > > Thanks! > Pablo > > On Fri, Jul 20, 2012 at 5:12 PM, Donna Dierker <[email protected]> > wrote: > > Sorry I'm not as familiar with that script, but I am familiar with scripts > > we use for importing human freesurfer files into Caret. Hopefully these > > excerpts might help: > > > > #Find c_ras offset between FreeSurfer surface and volume and generate > > matrix to transform surfaces > > MatrixX=`$FreeSurferInstallationDirectory/bin/mri_info --cras > > $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f1 -d' '` > > MatrixY=`$FreeSurferInstallationDirectory/bin/mri_info --cras > > $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f2 -d' '` > > MatrixZ=`$FreeSurferInstallationDirectory/bin/mri_info --cras > > $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f3 -d' '` > > Matrix1=`echo "1 0 0 ""$MatrixX"` > > Matrix2=`echo "0 1 0 ""$MatrixY"` > > Matrix3=`echo "0 0 1 ""$MatrixZ"` > > Matrix4=`echo "0 0 0 1"` > > MatrixCRAS=`echo "$Matrix1"" ""$Matrix2"" ""$Matrix3"" ""$Matrix4"` > > > > for Surface in white pial ; do > > caret_command -surface-apply-transformation-matrix > > $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere."$Surface"_orig.initial_mesh.coord.gii > > > > $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere.initial_mesh.topo.gii > > > > $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere."$Surface"_orig.initial_mesh.coord.gii > > -matrix $MatrixCRAS > > done > > > > Also, it might matter which volume you use (e.g., orig.mgz vs others). > > > > > > On Jul 20, 2012, at 3:55 PM, Pablo Polosecki wrote: > > > >> Hi Donna, > >> > >> I tried converting to RAS (this converst the heder information, which > >> apparently Caret does use) and the data matrix. The problem is the > >> same: the axes are correct (i.e., R points to the right side, D is > >> dorsal, etc). However, the surface and the volume still don't match in > >> Caret just like before. The issue seems to be some sort of offset in > >> the center of coordinates (when I click on the surface, the > >> highlighted points seem to be on a sensible medio-lateral and > >> dorso-ventral position. The problem is that the highlighted voxel is > >> too posterior in the volume view). Does this make sense or hint to a > >> possible solution to you perhaps? Worst case scenario, I could eyeball > >> how big this translation offset is, open the volume in matlab and have > >> it translate the data matrix. That way, it would roughly look okay, > >> but perhaps there is a more graceful way of handling this. > >> Thank you again, > >> Pablo > >> > >> On Fri, Jul 20, 2012 at 3:14 PM, Donna Dierker <[email protected]> > >> wrote: > >>> Be aware that what AFNI/Caret calls LPI, Freesurfer calls RAS. ;-) > >>> > >>> Can you try having mri_convert write the volume out as RAS, instead of > >>> LPI? Freesurfer will read the header and do the right flipping/rotating > >>> so that the surface aligns with it. Freesurfer did the right thing, as it > >>> understands that label. But that understanding is not what other software > >>> desires, in this case. (At least that's my read of your problem.) > >>> > >>> I think the axes don't need rotating anymore, but all three need flipping. > >>> > >>>> Dear List, > >>>> > >>>> I successfully completed the FS-to-F99 tutorial and now I am trying to > >>>> reproduce the steps with my own FreeSurfer dataset. According to the > >>>> FS-to-F99 tutorial, I need to have my mri volume "into LPI > >>>> orientation" and "the surfaces and volume need to be in register". My > >>>> original volume is in LIA orientation and is in register with my > >>>> surface, in the sense that when I open both on FreeSurfer's tkmedit > >>>> they are well-aligned to each other (I don't know how to check that > >>>> they are in register otherwise). If I use this surface and volume on > >>>> FS-to-F99's Stage-1 script and open the resulting spec file with > >>>> Caret, the volume's axes are wrong and the points on the surface and > >>>> in the volume don't match (I see this by clicking on the surface and > >>>> looking what is then highlighted on the volume window). If I transform > >>>> the volume into LPI orientation (using FreeSurfer's mri_convert > >>>> command), the new volume has the desired orientation and, > >>>> interestingly, it is still well-aligned to the surface in FreeSurfer's > >>>> tkmedit. However, when I use this surface and volume on the FS-toF99's > >>>> Stage-1 script, the volume and the surface are still not well-aligned, > >>>> even though the axes labels are now correct. How could I fix things so > >>>> that my volume and surface are in register when shown in Caret? > >>>> > >>>> Thank you very much, > >>>> Pablo > >>>> _______________________________________________ > >>>> 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 > > > > > > _______________________________________________ > > caret-users mailing list > > [email protected] > > http://brainvis.wustl.edu/mailman/listinfo/caret-users > > > > ---------- Forwarded message ---------- > From: David Van Essen <[email protected]> > To: "Caret, SureFit, and SuMS software users" <[email protected]> > Cc: > Date: Fri, 20 Jul 2012 20:25:22 -0500 > Subject: Re: [caret-users] Freesurfer Surface and Volume are not in register > when shown in Caret > Pablo, > > Registration to F99 should not be impacted by issues of scaling. (For > example, it doesn't affect our monkey-to-human registration despite large > scale differences.) But if you want to rescale just for the sake of > veridical representation, you can use > caret_command -surface-apply-transformation-matrix or Surface: Tranform: > Scale within Caret. > > David > > On Jul 20, 2012, at 6:27 PM, Pablo Polosecki wrote: > > > Hi Donna, > > > > Your excerpts were really inspiring. Many many thanks. I got them > > finally well-alligned. > > One more question: My volumes and surfaces have a nominal resolution > > of 1mm but they are actually .5mm (This means we scan at .5 mm > > resolution but we have to trick FreeSurfer into thinking we scan at > > 1mm because it doesn't display things correctly with other resolution > > values). Does this matter for working in Caret and registering to F99? > > Or do I have to rescale all my files back to .5mm? > > > > Thanks! > > Pablo > > > > On Fri, Jul 20, 2012 at 5:12 PM, Donna Dierker <[email protected]> > > wrote: > >> Sorry I'm not as familiar with that script, but I am familiar with scripts > >> we use for importing human freesurfer files into Caret. Hopefully these > >> excerpts might help: > >> > >> #Find c_ras offset between FreeSurfer surface and volume and generate > >> matrix to transform surfaces > >> MatrixX=`$FreeSurferInstallationDirectory/bin/mri_info --cras > >> $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f1 -d' '` > >> MatrixY=`$FreeSurferInstallationDirectory/bin/mri_info --cras > >> $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f2 -d' '` > >> MatrixZ=`$FreeSurferInstallationDirectory/bin/mri_info --cras > >> $InputFreesurferSubjectDirectory/mri/orig.mgz | cut -f3 -d' '` > >> Matrix1=`echo "1 0 0 ""$MatrixX"` > >> Matrix2=`echo "0 1 0 ""$MatrixY"` > >> Matrix3=`echo "0 0 1 ""$MatrixZ"` > >> Matrix4=`echo "0 0 0 1"` > >> MatrixCRAS=`echo "$Matrix1"" ""$Matrix2"" ""$Matrix3"" ""$Matrix4"` > >> > >> for Surface in white pial ; do > >> caret_command -surface-apply-transformation-matrix > >> $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere."$Surface"_orig.initial_mesh.coord.gii > >> $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere.initial_mesh.topo.gii > >> > >> $OutputRootDir/$Subject/$InitialMeshDirectory/$Subject.$Hemisphere."$Surface"_orig.initial_mesh.coord.gii > >> -matrix $MatrixCRAS > >> done > >> > >> Also, it might matter which volume you use (e.g., orig.mgz vs others). > >> > >> > >> On Jul 20, 2012, at 3:55 PM, Pablo Polosecki wrote: > >> > >>> Hi Donna, > >>> > >>> I tried converting to RAS (this converst the heder information, which > >>> apparently Caret does use) and the data matrix. The problem is the > >>> same: the axes are correct (i.e., R points to the right side, D is > >>> dorsal, etc). However, the surface and the volume still don't match in > >>> Caret just like before. The issue seems to be some sort of offset in > >>> the center of coordinates (when I click on the surface, the > >>> highlighted points seem to be on a sensible medio-lateral and > >>> dorso-ventral position. The problem is that the highlighted voxel is > >>> too posterior in the volume view). Does this make sense or hint to a > >>> possible solution to you perhaps? Worst case scenario, I could eyeball > >>> how big this translation offset is, open the volume in matlab and have > >>> it translate the data matrix. That way, it would roughly look okay, > >>> but perhaps there is a more graceful way of handling this. > >>> Thank you again, > >>> Pablo > >>> > >>> On Fri, Jul 20, 2012 at 3:14 PM, Donna Dierker <[email protected]> > >>> wrote: > >>>> Be aware that what AFNI/Caret calls LPI, Freesurfer calls RAS. ;-) > >>>> > >>>> Can you try having mri_convert write the volume out as RAS, instead of > >>>> LPI? Freesurfer will read the header and do the right flipping/rotating > >>>> so that the surface aligns with it. Freesurfer did the right thing, as > >>>> it > >>>> understands that label. But that understanding is not what other > >>>> software > >>>> desires, in this case. (At least that's my read of your problem.) > >>>> > >>>> I think the axes don't need rotating anymore, but all three need > >>>> flipping. > >>>> > >>>>> Dear List, > >>>>> > >>>>> I successfully completed the FS-to-F99 tutorial and now I am trying to > >>>>> reproduce the steps with my own FreeSurfer dataset. According to the > >>>>> FS-to-F99 tutorial, I need to have my mri volume "into LPI > >>>>> orientation" and "the surfaces and volume need to be in register". My > >>>>> original volume is in LIA orientation and is in register with my > >>>>> surface, in the sense that when I open both on FreeSurfer's tkmedit > >>>>> they are well-aligned to each other (I don't know how to check that > >>>>> they are in register otherwise). If I use this surface and volume on > >>>>> FS-to-F99's Stage-1 script and open the resulting spec file with > >>>>> Caret, the volume's axes are wrong and the points on the surface and > >>>>> in the volume don't match (I see this by clicking on the surface and > >>>>> looking what is then highlighted on the volume window). If I transform > >>>>> the volume into LPI orientation (using FreeSurfer's mri_convert > >>>>> command), the new volume has the desired orientation and, > >>>>> interestingly, it is still well-aligned to the surface in FreeSurfer's > >>>>> tkmedit. However, when I use this surface and volume on the FS-toF99's > >>>>> Stage-1 script, the volume and the surface are still not well-aligned, > >>>>> even though the axes labels are now correct. How could I fix things so > >>>>> that my volume and surface are in register when shown in Caret? > >>>>> > >>>>> Thank you very much, > >>>>> Pablo > >>>>> _______________________________________________ > >>>>> 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 > >> > >> > >> _______________________________________________ > >> 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 > > > > _______________________________________________ > 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
