Re: [caret-users] Generating cortical hull surface

[Rosita Shishegar]

Thanks Donna!

On Fri, Jan 15, 2016 at 3:07 AM, Donna Dierker 
wrote:

> Hi Rosita,
>
> It can be so frustrating.  Sounds like you have some good alternatives to
> explore.
>
> This link details one user's alignment issues and the solution:
>
> http://brainmap.wustl.edu/pub/donna/IN/JOHN/align.html
> login pub
> password download
>
>
> Donna
>
>
> On Jan 13, 2016, at 10:09 PM, Rosita Shishegar <
> r.shishe...@student.unimelb.edu.au> wrote:
>
> > Hi Donna,
> >
> > Thanks for the explanation.
> >
> > Unfortunately, as a result of preprocessing steps my surfaces are
> transformed, resized, and they are not even aligned with the original
> volumes anymore and I prefer not to bother with registration with original
> data.
> >
> > The final goal was to compute sulcal depth. However, computing the
> segmentation volume  from the surface would serve the purpose for me
> because I have written the codes for the rest of the steps. As you
> suggested, I would try to see what else may work for my sheep brains.
> >
> > Thanks again for all the help.
> >
> > Cheers,
> > Rosita
> >
> > On Tue, Jan 12, 2016 at 3:41 AM, Donna Dierker 
> wrote:
> > Hi Rosita,
> >
> > None of the parameters in these commands makes sense for sheep:
> >
> > > caret_command -volume-create 176 208 176 hull_vol.nii
> > volume dimensions
> > > caret_command -volume-set-origin hull_vol.nii hull_vol.nii -88.0
> -123.0 -75.0
> > volume origin
> > > caret_command -volume-set-spacing hull_vol.nii hull_vol.nii 1 1 1
> > voxel dimensions
> >
> > How did you get that surface?  Typically, you start off with a T1 or T2
> volume and segment it using some software.  Then the binary segmentation
> volume is tesselated into a surface.  You already had the surface, so did a
> volume give rise to that?  If so, its dimensions, origin, and voxdims give
> clues to what these should be.  In particular, if there is some volume you
> want this surface to align with, that might be a good one to use for these
> parameters.
> >
> > > caret_command -surface-to-segmentation-volume
> sheep_90_rh.pial.surf.coord
> > > sheep_90_rh.pial.surf.topo hull_vol.nii
> >
> > The above command generates a ribbon of voxels along the surface, and
> I'm not sure exactly how thick.  You can specify inner/outer thickness in
> the Caret GUI (Surface: Region of Interest: Select all nodes: Create Volume
> Region of Interest).  This is more or less what the above caret_command
> does, but it uses a default inner/outer thickness that might be too thick
> for your sheep.
> >
> > Again, not sure what the end game is here (e.g., do you want sulcal
> depth? hull? registration?).  I fear there will be other showstopper steps
> down the road that are more sheep-proof. ;-)
> >
> > (Sorry, Tim, I'm having trouble thinking of bd puns here.)
> >
> > Donna
> >
> >
> > On Jan 10, 2016, at 9:31 PM, Rosita Shishegar <
> r.shishe...@student.unimelb.edu.au> wrote:
> >
> > > Hi Donna,
> > >
> > > Thanks a lot for the help.
> > >
> > > I ran the commands as you suggested and now I can compute the volume
> and the hull. My understanding is that the inputs that you chose for
> creating volume (-volume-create 176 208 176  ) and the origin
> (-volume-set-origin hull_vol.nii hull_vol.nii -88.0 -123.0 -75.0) were the
> reason I was getting error. Should I use these numbers for any data?
> > >
> > > As you mentioned, since my dataset includes sheep brains, computed
> cerebral hull may not be as convex as it should be.  However, if I can
> compute an accurate segmentation volume with Caret, I can compute the hull
> with more adjusted tuning parameters in Matlable.
> > >
> > > The problem is that the brains are smaller than human brain and and I
> think volume spacing 1 1 1 is quite big for them. I tried to use a smaller
> volume spacing (e.g. caret_command -volume-set-spacing hull_vol.nii
> hull_vol.nii 0.2 0.2 0.2) but it is giving me errors. Do you think there
> would be any way to use volume spacing smaller than 1 for this caret
> command.
> > >
> > > Cheers,
> > > Rosita
> > >
> > >
> > >
> > > On Sat, Jan 9, 2016 at 10:16 AM,  wrote:
> > > Hi Rosita,
> > >
> > > This is a sheep brain, so I don't know that the hull Caret generates
> will
> > > be suitable for your purposes, because this was designed for human
> brains.
> > >
> > > Still, I tried these steps after converting to caret coord/topo:
> > >
> > > caret_command -volume-create 176 208 176 hull_vol.nii
> > > caret_command -volume-set-origin hull_vol.nii hull_vol.nii -88.0
> -123.0 -75.0
> > > caret_command -volume-set-spacing hull_vol.nii hull_vol.nii 1 1 1
> > > caret_command -surface-to-segmentation-volume
> sheep_90_rh.pial.surf.coord
> > > sheep_90_rh.pial.surf.topo hull_vol.nii
> > > caret_command -surface-identify-sulci Other.sheep.R.1002.spec right
> > > hull_vol.nii sheep_90_rh.pial.surf.topo sheep_90_rh.pial.surf.coord
> > > sheep_90_rh.pial.surf.coord NIFTI_GZIP
> > >
> 

Re: [caret-users] Generating cortical hull surface

[Donna Dierker]

Hi Rosita,

It can be so frustrating.  Sounds like you have some good alternatives to 
explore.

This link details one user's alignment issues and the solution:

http://brainmap.wustl.edu/pub/donna/IN/JOHN/align.html
login pub
password download


Donna


On Jan 13, 2016, at 10:09 PM, Rosita Shishegar 
 wrote:

> Hi Donna,
> 
> Thanks for the explanation.
> 
> Unfortunately, as a result of preprocessing steps my surfaces are 
> transformed, resized, and they are not even aligned with the original volumes 
> anymore and I prefer not to bother with registration with original data.
> 
> The final goal was to compute sulcal depth. However, computing the 
> segmentation volume  from the surface would serve the purpose for me because 
> I have written the codes for the rest of the steps. As you suggested, I would 
> try to see what else may work for my sheep brains. 
> 
> Thanks again for all the help.
> 
> Cheers,
> Rosita
> 
> On Tue, Jan 12, 2016 at 3:41 AM, Donna Dierker  
> wrote:
> Hi Rosita,
> 
> None of the parameters in these commands makes sense for sheep:
> 
> > caret_command -volume-create 176 208 176 hull_vol.nii
> volume dimensions
> > caret_command -volume-set-origin hull_vol.nii hull_vol.nii -88.0 -123.0 
> > -75.0
> volume origin
> > caret_command -volume-set-spacing hull_vol.nii hull_vol.nii 1 1 1
> voxel dimensions
> 
> How did you get that surface?  Typically, you start off with a T1 or T2 
> volume and segment it using some software.  Then the binary segmentation 
> volume is tesselated into a surface.  You already had the surface, so did a 
> volume give rise to that?  If so, its dimensions, origin, and voxdims give 
> clues to what these should be.  In particular, if there is some volume you 
> want this surface to align with, that might be a good one to use for these 
> parameters.
> 
> > caret_command -surface-to-segmentation-volume sheep_90_rh.pial.surf.coord
> > sheep_90_rh.pial.surf.topo hull_vol.nii
> 
> The above command generates a ribbon of voxels along the surface, and I'm not 
> sure exactly how thick.  You can specify inner/outer thickness in the Caret 
> GUI (Surface: Region of Interest: Select all nodes: Create Volume Region of 
> Interest).  This is more or less what the above caret_command does, but it 
> uses a default inner/outer thickness that might be too thick for your sheep.
> 
> Again, not sure what the end game is here (e.g., do you want sulcal depth? 
> hull? registration?).  I fear there will be other showstopper steps down the 
> road that are more sheep-proof. ;-)
> 
> (Sorry, Tim, I'm having trouble thinking of bd puns here.)
> 
> Donna
> 
> 
> On Jan 10, 2016, at 9:31 PM, Rosita Shishegar 
>  wrote:
> 
> > Hi Donna,
> >
> > Thanks a lot for the help.
> >
> > I ran the commands as you suggested and now I can compute the volume and 
> > the hull. My understanding is that the inputs that you chose for creating 
> > volume (-volume-create 176 208 176  ) and the origin (-volume-set-origin 
> > hull_vol.nii hull_vol.nii -88.0 -123.0 -75.0) were the reason I was getting 
> > error. Should I use these numbers for any data?
> >
> > As you mentioned, since my dataset includes sheep brains, computed cerebral 
> > hull may not be as convex as it should be.  However, if I can compute an 
> > accurate segmentation volume with Caret, I can compute the hull with more 
> > adjusted tuning parameters in Matlable.
> >
> > The problem is that the brains are smaller than human brain and and I think 
> > volume spacing 1 1 1 is quite big for them. I tried to use a smaller volume 
> > spacing (e.g. caret_command -volume-set-spacing hull_vol.nii hull_vol.nii 
> > 0.2 0.2 0.2) but it is giving me errors. Do you think there would be any 
> > way to use volume spacing smaller than 1 for this caret command.
> >
> > Cheers,
> > Rosita
> >
> >
> >
> > On Sat, Jan 9, 2016 at 10:16 AM,  wrote:
> > Hi Rosita,
> >
> > This is a sheep brain, so I don't know that the hull Caret generates will
> > be suitable for your purposes, because this was designed for human brains.
> >
> > Still, I tried these steps after converting to caret coord/topo:
> >
> > caret_command -volume-create 176 208 176 hull_vol.nii
> > caret_command -volume-set-origin hull_vol.nii hull_vol.nii -88.0 -123.0 
> > -75.0
> > caret_command -volume-set-spacing hull_vol.nii hull_vol.nii 1 1 1
> > caret_command -surface-to-segmentation-volume sheep_90_rh.pial.surf.coord
> > sheep_90_rh.pial.surf.topo hull_vol.nii
> > caret_command -surface-identify-sulci Other.sheep.R.1002.spec right
> > hull_vol.nii sheep_90_rh.pial.surf.topo sheep_90_rh.pial.surf.coord
> > sheep_90_rh.pial.surf.coord NIFTI_GZIP
> >
> > And I got a cerebral hull volume, but it probably isn't as convex as you'd
> > like.  How are you using this?
> >
> > Donna
> >
> >
> > > Thanks for offering help Donna!
> > >
> > > I uploaded the surface