[Insight-users] modification of itkBinaryBallStructuringElement

[Irene Gonzálvez García]

Dear all,
I've successfully run a dilatation using a BallStructuringElement in  
4D. Now, I am trying to modify the structuring element. What I would 
like to create is an asymmetric BallStructuringElement. Making an 
example with a 3x3 2D structuring element, i.e. in chapter 11 page 722, 
iterators, I would like to make zero the pixels 0, 3 and 6. In the 
itkBinaryBallStructuringElement.txx, I've modified part of the code:

 // Iterate through the entire image and set interior pixels to 1
   for(sfi.GoToBegin(); (!sfi.IsAtEnd()); ++sfi)
    {
     sfi.Set(NumericTraits<TPixel>::One);
     }
//Set pixels 0,3 and 6 equal to 0
   for(sfi.GoToBegin(); (!sfi.IsAtEnd()); ++sfi)
    {
    sfi.Set(NumericTraits<TPixel>::Zero);
   ++sfi;   //I write this 2 lines because the operatos += is not 
defined, so it isn't possible to write in the for definition sfi+=3;
   ++sfi;
   }


Working in 2D, the pixels 0,3 and 6 should be equal to zero and the rest 
equal to 1, but the result of the dilation is not the expected one, so I 
think the problem is in this part of the code because the dilatation 
algorithm works fine. Is it the correct way to set pixels equal to 
zero?. Continuing with this example and for a 4D image I would like to 
create a structuring element in 4D, making it symmetric for (X,Y,Z) and 
asymmetric the temporal dimension (from the center to the left side 
equal to zero and from the center to the right equal to 1(including the 
center)). How can I identify pixels in 4D?

Best regards and thank you for your help
Irene