[Insight-users] Need advice on LV MRI segmentation

[Barry Branham]

Hi Luis,

I just resurrected my website using Drupal this afternoon.  It was 
either that or facebook.
I put the Dicom file and two screen shots at 
http://branham.ws/?q=Navigation/4

I revisited BalloonForceFilter, Luis, and discovered you were actually 
right the first time in saying it was for 3D.  It's just the description 
that is wrong.  It is in fact BalloonForce3DFilter.  The reason it was 
crashing for me was I was giving it a 2D Mesh and it was trying to 
access the 3rd dim components.  The test app you mentioned 
"itk2DDeformableTest.cxx", uses Mesh<double> with the default Dimension 
parameter == 3 and then uses SphereMeshSource to initialize the mesh.  
So itk2DDeformableTest.cxx should be renamed and the description in 
itkBalloonForceFilter.h needs updating.

Looking forward to any ideas on segmenting the Epi.

Thanks,
Barry



Luis Ibanez wrote:
> Hi Barry,
>
> Could you please post screenshots of your images to a public web site ?
>
> In particular, the original tagged MRI image, and the segmentation that
> you are obtaining so far.
>
> If we can see the images, it will be a lot easier to advise you on the
> potential segmentation methods that you may want to try.
>
>      Regards,
>
>
>             Luis
>
>
> --------------------
> On Sun, Nov 8, 2009 at 9:36 PM, Barry Branham <bhb at branham.ws> wrote:
>   
>> Hi all,
>>
>> I'm working on a project to segment left ventricular walls from 2D MRI
>> images and would like to get advice on how to proceed with the epicardial
>> walls.
>>
>> My project is to segment left ventricular walls (the round ones) from 2D MRI
>> images.  I'm using 2D because the main application uses Tagged MRI images
>> for determining tissue displacements for input into an FEA application for
>> determining wall stress/strains.  And the Imaging sequence requires long
>> patient breath holds so apparently they are limited in the number of slices
>> output so my slice spacing is about 8mm.  Fortunately they have recently
>> also acquired non-tagged images that I can use for segmentation.  I have had
>> good luck segmenting the endocardial walls using IntensityWindow ->
>> CurvatureFlow -> ConfidenceConnected, then
>> DeformableMesh2DFilter which I converted from DeformableMesh3DFilter.
>>
>> The epicardial walls are harder because of the lack of large gradients on
>> part of the outside (non-septal) wall.  My current plan is to use a
>> deformable model generated by getting the major axis from the endo wall and
>> generating a slightly larger oriented ellipse from that.  I need a good way
>> to enlarge it to fit the epicardial wall.
>>
>> Thanks for any suggestions,
>> Barry Branham
>> Cardiothoracic Surgery Research
>> Washington University School of Medicine
>> St. Louis, MO
>>
>> _____________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Kitware offers ITK Training Courses, for more information visit:
>> http://www.kitware.com/products/protraining.html
>>
>> Please keep messages on-topic and check the ITK FAQ at:
>> http://www.itk.org/Wiki/ITK_FAQ
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.itk.org/mailman/listinfo/insight-users
>>
>>     
>
>