Hi Donna,
Thank you, that clears up a lot. Just to clarify, my main purpose is to use
Caret to generate sulcal depth maps for each subject and run statistical
analysis (such as ANOVA) to compare the results of two groups. I have
completed importing all my data to the fs_LR standard mesh, and now I have
to generate sulcal depth maps and run analyses. To do this, I am planning
on running the "generate depth" function from PALS-B12/preborder.sh. Will
this generate the appropriate input parameters for running the one-way
ANOVA test through caret_command?
Here is a list of what's needed:
METRIC STATISTICS ONE-WAY ANOVA
caret_command -metric-statistics-anova-one-way
<fiducial-coord-file>
<open-topo-file>
<distortion-metric-shape-file>
<distortion-column-number>
<output-file-names-prefix>
...
<metric-file-names>
After the import of the data into fs_LR mesh, I have multiple coord files,
shape files, and closed topo files. I do not have specific paint or border
files. So I am not exactly sure how to exclude the paint files that label
the medial wall from my analysis, like you stated previously. I found a
command in postborder.sh that converts the closed topo file into an open
topo file using the roi file that's generated for the medial wall. Any
guidance on doing that would be appreciated.
Please let me know if I'm going on the right track to generate the sulcal
depth maps and run my analyses.
Thank you,
Eshita
On Nov 15, 2013, at 10:07 AM, Donna Dierker <[email protected]> wrote:
>Hi Eshita,
>
>topi = topology, and I hate auto-correct
>You mean the surface_shape/metric files -- not the topology, right?
>
>When all your data is on a standard mesh, as Caret requires to do group
>analysis (except for summary stats that are not vertex-wise, e.g., gyrification
>index), then typically the stats tests use a single mean midthickness and open
>topology for computing the areas that are used for the TFCE/cluster
>distributions.
>
>You typically don't need to generate an open topo file for each subject on the
>standard mesh. You have a single paint/label file that labels the medial wall
>vertices, so you can exclude them from your analysis. Or in this case, you
>have an open topology file that will accomplish the same thing. If they're all
>on standard mesh, then your standard open topi file should look reasonable with
>all your standard mesh midthickness surfaces.
>
>An example of a time when you would need an open topi is when you want the
>gyrification index computed on the native mesh. Then you can use a strategy
>like the PALS_B12.LR/postborder.sh
>(http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2011_10/PALS_B12.LR/postborder.sh
> ; login pub, password download) uses in the freesurfer to PALS pipeline, which
>essentially writes a border around the standard mesh medial wall; "unprojects"
>the border on the standard mesh midthickness; and then projects it on the
>native mid thickness. When you do that, you need to project to/from the
>spherical or ellipsoid, so that your medial wall border points don't get hosed.
>
>You can use your own study-specific mean midthickness, but then you should
>compute your own mean distortion metric to go with it. Besides being a bit of
>a hassle, it strikes me that your results are slightly less comparable to
>others computed on the fs_LR standard mesh that used the more standard Conte69
>mean mid thickness/distortion. I don't use study-specific files for this
>purpose, unless the populations are so different (e.g., baby vs adult) that you
>can't use the adult.
>
>Visualization is another story. I almost always show the study-specific mean
>midthickness for each group in morphometry studies.
>
>Donna
On Thu, Nov 14, 2013 at 1:38 PM, Eshita Shah <[email protected]> wrote:
> Hi Donna,
>
>
> Thank you for your help. From what I understand, the topology files from each
> subject have to be compiled (using caret_command -metric-composite) into one,
> and that file is then entered into the ANOVA test. My question however, is
> regarding the generation of the open topo files for each subject. I have used
> freesurfer_to_fs_LR pipeline, and I'm wondering if the topo files generated
> (there is only one for each subject) are indeed the open topo file, because
> looking at the script for freesurfer_to_fs_LR it looks like it is the closed
> topo file. Any advice on how to obtain an open topo file?
>
>
> Also, to clarify, the first parameter in the ANOVA test is a fiducial coord
> file. I have used caret_command -surface-average to generate an average
> fiducial file for each hemisphere. Is this correct?
>
>
> Please let me know.
>
>
> Thanks,
>
> Eshita
>
>
> On Wed, Nov 13, 2013 at 10:50 PM. Donna Dierker wrote:
>
> >Hi Eshita,
>
> >
> >I always use the open topology for this purpose (i.e., excludes only medial
> >wall vertices). The files here will be helpful:
>
> >http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2013_11/TFCE_164k/
> >login pub
> >password download
> >
> >You can get them all in this zip file:
> >
> >http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2013_11/TFCE_164k.zip
> >
> >Just to explain what is going on, the areas in the TFCE/cluster computations
> >are computed on the Conte69 mean mid thickness, with the open topo file
> >(excluding medial wall). The distortion maps pump up the areal value where
> >substantial smoothing occurs as a result of averaging individuals' coordinate
> >files (e.g., high 3D variability). The intent is to make the areas more like
> >an individual's area would be in that region. For folks not attuned to what
> >you are doing, this is during group analysis.
>
> >
> >Cheers,
> >
> >Donna
>
> >
> >On Tue, Nov 12, 2013 at 1:18 PM, Eshita Shah <[email protected]> wrote:
>
>> Hello,
>>
>> Thanks for your input. I successfully was able to use freesurfer_to_fs_LR
>> Pipeline to import my FreeSurfer files into caret, however when I try
>> running ANOVA, it asks for certain files that have not been generated by
>> the pipeline. Specifically, how do I generate the
>> "distortion-metric-shape-file" that is being asked for? Lastly, is the
>> .topo file that is generated via the pipeline the open topo file or closed?
>> Previously I was able generate the closed topo file, so I'm not sure if the
>> freesurfer_to_fs_LR pipeline does the same. The parameter required for the
>> ANOVA analysis is the open topo file.
>>
>> Thank you,
>> Eshita Shah
>>
>>
>> On Thu, Nov 7, 2013 at 4:12 PM, Rouhollah Abdollahi
>> <[email protected]>wrote:
>>
>>> Hi
>>> Actually the code import the original data from freesurfer to caret then
>>> automatically you will have different node number for different subjects
>>> and hemispheres. To have the same mesh you can use Freesurfer_to_fs_LR
>>> Pipeline which is available in the caret website. It imports all the data
>>> to the same mesh which here is fs_LR mesh.
>>> Hope it helps
>>> Best
>>> Rouhi
>>> On Nov 8, 2013 12:06 AM, "Eshita Shah" <[email protected]> wrote:
>>>
>>>> Hello,
>>>>
>>>> I have just recently started using Caret, and I am running the
>>>> freesurfer2caret.sh script in order to import my FreeSurfer files into
>>>> Caret as well as generate sulcal depth for all subjects. I tried doing a
>>>> One-Way ANOVA test, but I've realized that the number of nodes in the
>>>> metric/surface_shape files for the two subjects are different. How is it
>>>> possible that the same script is creating files with different node
>>>> numbers? Also, within each subject, sometimes the node numbers for the left
>>>> and right hemisphere are different as well. How can I resolve this issue so
>>>> I can successfully run ANOVA on my subjects?
>>>>
>>>> Any help would be appreciated.
>>>>
>>>> Thank you,
>>>> Eshita Shah
>>>>
>>>> _______________________________________________
>>>> 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
>>>
>>
>>
--
Eshita Shah
University of California, Los Angeles | 2014
B.S. Neuroscience
[email protected]
_______________________________________________
caret-users mailing list
[email protected]
http://brainvis.wustl.edu/mailman/listinfo/caret-users
