[caret-users] draw borders for spherical registration - Expediting the process

Fri Jul 6 16:53:05 CDT 2007

Hi Julia, plus others interested in registering surfaces (FreeSurfer-  
as well as SureFit-generated) to the PALS human atlas,

John and Donna have already provided useful technical comments that  
have hopefully addressed the specific questions you raised.

This email provides a heads-up on two parallel efforts we are  
currently undertaking to expedite surface-based registration,  
particularly mapping FreeSurfer surfaces onto PALS.

1) A revised process will bypass flat maps altogether and instead  
draw the 'Core 6' landmark contours directly on inflated surfaces.   
The requisite tools are already in place; we're currently revising  
the instructions appropriately.  This should save a fair amount of time.

2) A more ambitious effort is to fully automate the process of  
drawing landmark contours.  There are many steps to doing this with  
high fidelity, but what we've implemented so far look quite encouraging.

The objective is to allow batch process mapping of FreeSurfer to PALS  
- fully automatically, but with quality control checks included.

David VE

On Jul 5, 2007, at 9:30 AM, Donna Dierker wrote:

> Hi Julia,
>
> On your initial flat map, none of the red patches along the edges  
> would worry me, but we would check the ones I've circled in the  
> attached capture.  By check, I mean click on the red patch in the  
> main window, but use an inflated view in Window 2 to locate the  
> crossover region in the inflated surface.  The patch in the upper  
> left may end up being nothing, but the one in the lower right looks  
> like a handle connecting cortical regions that are separated by a  
> significant geodesic distance.
>
> If you're processing enough subjects that a case or two like this  
> is down in the noise, then this isn't a show stopper.  But this is  
> something we'd fix in the segmentation and regenerate the surface,  
> for one of our own studies.
>
> Donna
>
> On 07/03/2007 02:01 PM, John Harwell wrote:
>> Hi Julia,
>>
>> It is not unusual for there a few red highlights around the edges  
>> at this stage of the flattening process.  You could continue the  
>> flattening process and it might turn out okay.
>>
>> If there are no topological problems with the original 3D fiducial  
>> surface, then all surfaces derived from it (spherical, etc) should  
>> be topologically correct.  Load the original 3D surface into  
>> Caret's main window and select Surface Menu->Topology->Topology  
>> Error Report (it may take a minute to execute).  If the "Number of  
>> Handles" is zero, then the 3D surface and the "derived" surfaces  
>> (spherical, flat) are topologically correct and you can proceed  
>> with flattening.
>>
>> If the 3D surface does contain handles, the best solution is to go  
>> back to the original segmentation volume and make corrections.   
>> However, since your surface was not generated within Caret, there  
>> is a new feature in Caret that may help.  With the original 3D  
>> surface in Caret's main window, Select Surface Menu->Topology- 
>> >Correct Fiducial Surface Topology (Remove Handles).  This process  
>> will attempt to re-tessellate the surface in order to remove any  
>> handles and will create new coordinate and topology files that  
>> will need to be saved.  This is a new process, so examine the  
>> corrected surface to verify that the process has not done anything  
>> bad.  If things look good, use the new coordinate and topology  
>> file and restart the flattening process.
>>
>> When I have time, I will modify Caret so that the beginning of the  
>> flattening process will check for topological problems.
>>
>> ----------------------------------------------------------
>> John Harwell
>> john at brainvis.wustl.edu
>> 314-362-3467
>>
>> Department of Anatomy and Neurobiology
>> Washington University School of Medicine
>> 660 S. Euclid Ave.    Box 8108
>> St. Louis, MO 63110   USA
>>
>> On Jul 3, 2007, at 10:22 AM, Julia Molony wrote:
>>
>>> John,
>>>
>>> Your answer was very helpful.  Now I'm stuck with an initial flat  
>>> map that has red highlights on a few of the edges.  The tutorial  
>>> says this means there are handles that need to be corrected in  
>>> the segmentation volume.  I don't have a volume.  I have  
>>> Freesurfer surfaces that I'm using to compare warping methods.   
>>> Is there any solution to this cross over correction problem for  
>>> already created surfaces?
>>>
>>> There's just a little bit of red.  Here's a picture.  I can send  
>>> more information, if that would be helpful.
>>>
>>> Thank you.
>>>
>>> Julia
>>>
>>> -- 
>>> Julia Molony
>>> SSCC/NIMH/NIH
>>>
>>> <Init.Surf.crossovers 5.5_Figure40.jpg>
>>>
>>>
>>>
>>>
>>> On Jul 2, 2007, at 12:21 PM, John Harwell wrote:
>>>
>>>>
>>>> Hi Julia,
>>>>
>>>> During the flattening process, the medial wall border is  
>>>> normally drawn on the Compressed Medial Wall Surface.  The CMW  
>>>> surface is a sphere with the medial portion made flat so that 2D  
>>>> borders can be drawn on it.  This surface should be selected  
>>>> from the model selection control in the main window's toolbar.
>>>>
>>>> The "Smooth Fiducial Medial Wall" option results in the fiducial  
>>>> surface's medial wall being smoothed during the flattening  
>>>> process.  Smoothing will change the shape of the medial wall so  
>>>> that it is relatively flat.  This option does not affect the  
>>>> sphere.  There is normally some distortion around on the sphere  
>>>> that should be removed by the morphing process that is part of  
>>>> flattening.
>>>>
>>>> ----------------------------------------------------------
>>>> John Harwell
>>>> john at brainvis.wustl.edu
>>>> 314-362-3467
>>>>
>>>> Department of Anatomy and Neurobiology
>>>> Washington University School of Medicine
>>>> 660 S. Euclid Ave.    Box 8108
>>>> St. Louis, MO 63110   USA
>>>>
>>>> On Jul 2, 2007, at 10:09 AM, Julia Molony wrote:
>>>>
>>>>> I'm attempting to go through Caret's spherical registration  
>>>>> stuff but first I need to work out how to draw all the  
>>>>> borders.  I'm following the 5.5 tutorial and I'm stuck trying  
>>>>> to draw the medial wall borders.  When I open the draw borders  
>>>>> dialogue, choose the border name, set type to closed and  
>>>>> dimension to 2D, and press apply, I can't draw the border.  I  
>>>>> click on the sphere and hold the left mouse button and nothing  
>>>>> is drawn.  Any ideas on what's going on?  I checked the D/C to  
>>>>> see if "show borders" was selected and it is.  I've drawn  
>>>>> borders in the past with no trouble, so I'm not sure what's  
>>>>> happening here.
>>>>>
>>>>> Also, what does it mean in the "Flatten Full or Partial  
>>>>> Hemisphere" box when it says "Smooth Fiducial Medial Wall".  It  
>>>>> looks like this smoothing is causing distortions in the  
>>>>> sphere.  It makes the sphere look sort of swirly around the  
>>>>> medial wall area.
>>>>>
>>>>> Julia
>>>>>
>>>>>
>>>>> -- 
>>>>> Julia Molony
>>>>> SSCC/NIMH/NIH
>>>>>
>
> <Init.Surf.crossovers_check.jpg>
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://pulvinar.wustl.edu/mailman/listinfo/caret-users