[Insight-users] ConfidenceImageFilter

[Luis Ibanez]

Hi Samuel,

As suggested, a method was added to the
ConfidenceConnectedImageFilter that allows
to define the size of the initial neighborhood
over which the mean and variance around the
seed point will be computed.

Now you can invoke:

  confidenceFilter->SetInitialNeighborhoodRadius( 5 );

The default value is still=1.


You can in principle adventure with moderate values
of the radius (lets say 10~20 pixels). This shouldn't
have much of a performance hit since the neighborhood
is used only once in the initial evaluation aroud the
seed.

In fact, using a large radius should allow you to
reduce the number of iterations required for the
filter. Just make sure that whatever the size of the
neighborhood is, it is fully contained inside the
region you are intended to segment.

This should make the algorithm less sensitive to
the initial guess of the seed point.

--

Additionally, the

     - MeanImageFunction
     - VarianceImageFunction

have now methods:

      SetNeighborhoodRadius()


Please let us know if you have further questions.

Thanks


     Luis

===============================================

Samuel Rodríguez Bescos wrote:

> Thanks for the information.
> 
> Sam
> ----- Original Message -----
> From: "Luis Ibanez" <luis.ibanez@kitware.com>
> 
>>Hi Samuel,

>>

.......

>>
>>Probably we may want to add a method:
>>
>>          void SetInitialNeigborhoodRadius(int)
>>
>>to the itk::ConfidenceConnectedImageFilter.
>>
>>>
>>>
>>>I read the information at the website but I have a question. How are the
>>>pixels at the neighborhood seed evualuated in the first iteration?. What
>>>are the mean an variance values?.
>>>
>>>
>>>
>>>Thanks in advance,
>>>
>>>
>>>
>>>Sam
>>>
>>>
>>
>>
>>
> 
>