[Insight-users] Active Shape Model

[Lydia Ng]

Hi Steven,

I believe there is another group in the consortium aiming at integrating
active shape models to ITK. So they might have more direct advice - here
I will just give my 2 cents worth ....

>   I have built a statistical model of the cartilage from a training
set of
> manual segmentations using PCA. (without ITK - was there a way with
ITK?)

You could use the vnl_generalized_eigensystem to compute the PCA.

itkImagePCAShapeModelEstimator uses this inside for computing PCA of
images - but this assume a signed distance representation of the shape.
For a different shape description, you will most likely to have to
create a new class that does something similar.

> Now I
> am wondering how ITK can help me in the search for the model within a
new
> example. Cootes describes a Multi-Resolution ASM search algorithm,
which I
> would
> like to adapt to ITK somehow. Should I be looking at Model Based
> Registration?
> If so what will allow me to model the variation of the shape? Could
> someone
> please give me a hint on the ITK structures I should use?

If you can reformulated your problem as a general optimization problem
then you can use the ITK Optimizer paradigm. This is what is done for
most of the image to image registration algorithms.

You would then need to define some sort of controller class like
ImageRegistrationMethod, ImageToSpatialObjectRegistrationMethod or=20
MultiResolutionImageRegistrationMethod.


Hope this helps,
- Lydia
=20

> -----Original Message-----
> From: sdor5151 at mail . usyd . edu . au [mailto:sdor5151 at mail . usyd . edu . au]
> Sent: Friday, September 26, 2003 12:06 AM
> To: insight-users at itk . org
> Subject: [Insight-users] Active Shape Model
>=20
>=20
> Dear itk users,
>=20
>   I am writing an implementation of the Active Shape Model for my
thesis
> and
> would like to apply it to segmentation of the knee cartilage
boundaries.
>=20
>   I have built a statistical model of the cartilage from a training
set of
> manual segmentations using PCA. (without ITK - was there a way with
ITK?)
> Now I
> am wondering how ITK can help me in the search for the model within a
new
> example. Cootes describes a Multi-Resolution ASM search algorithm,
which I
> would
> like to adapt to ITK somehow. Should I be looking at Model Based
> Registration?
> If so what will allow me to model the variation of the shape? Could
> someone
> please give me a hint on the ITK structures I should use?
>=20
>   Any help would be greatly appreciated,
>=20
>   Kind Regards,
>=20
> Steven
>=20
> -------------------------------------------------
> This mail sent through IMP: www-mail.usyd.edu.au
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk . org
> http://www . itk . org/mailman/listinfo/insight-users