[caret-users] RadialPositionMap file

[Donna Dierker]

Hi Traci,

You have residual problems with the Freesurfer surfaces?  How old are they?

I doubt there is anything comparable to the RadialPositionMap (RPM) file 
in the Freesurfer subject's directory.  (The radial position map is sort 
of a probabilistic version of the segmentation; the RPM gets thresholded 
to generate the segmentation.  Both the RPM and Caret segmentation 
target the midthickness, rather than WM/GM or pial boundaries.  While it 
is easy to generate a midthickness surface, it is not so easy to 
generate a segmentation volume from the white/pial volumes.  You can 
fill the midthickness surface, as we do to generate a depth map from 
Freesurfer surfaces, but this segmentation is not as high quality as a 
normal Caret segmentation would be.  And this has nothing to do with 
Freesurfer; on the contrary, it is just the fact that it is being 
back-generated from the surface, rather than the surface being generated 
from the segmentation.)

I would definitely NOT recommend trying to correct the segmentation 
generated by the freesurfer2caret.sh script, for the reasons explained 
above.  And since it isn't clear to me how to generate a high quality 
midthickness segmentation from either the Freesurfer surface or the 
Freesurfer volumes, I think your problems are bigger than a missing RPM.

You might be able to import the wm.mgz and ?h.ribbon.mgz into Caret; 
fill the latter; and run just the Graph error correction on one/both of 
them.  But you'd need to inject the results back into the Freesurfer 
processing streams at the right times/calls, so that the results 
percolate to your surfaces, and I just don't know the processing stream 
well enough to help you there.

The graph topology correction algorithm was not developed here, and 
perhaps a free binary is available.  I'm guessing it doesn't read mgz, 
but mri_convert has several output options.

Sorry I can't think of an easier way.  If it's any comfort, I find 
myself wanting a Freesurfer midthickness segmentation often -- not so 
much for topology correction; mine generally don't need that, but for 
other purposes.

The only thing I would add is that if the topological errors are large 
enough to confound registration, then they could affect many upstream 
Freesurfer data files, statistics, and other outputs (i.e., consider 
fixing them in Freesurfer and re-running recon-all).

Donna

On 02/27/2009 09:18 AM, Sandoval, Traci I wrote:
>
> I am using Freesurfer recon all script to create my surfaces, ie. 
> Inflated, spherical, pial, white etc. Then I am  using a caret script 
> to add the freesurfer surfaces to a caret spec file, create 
> midthickness file and create a segmentation file. I would like to use 
> caret to clean up the surfaces but when I run the second part of the 
> segmentation process I am missing a RadialPositionMap file. Can I 
> create this file from a caret_command, or is there a comparable 
> freesurfer file I can use, or can I create it through the caret gui? 
> Thank you. Traci
>
>  
>
> ___________________              _
>
> Traci Sandoval
> Research Assistant
> NeuroPsychometric Research Lab
> School of Behavioral and Brain Sciences - University of Texas at Dallas
> Center for BrainHealth - University of Texas at Dallas
> Department of Psychiatry - University of Texas Southwestern Medical Center
>
>  
>
>  
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>