[Insight-users] VED filter multi-scale parameters question

[Oleksandr Dzyubak]

Dear ITKers,

I just downloaded the VED filter by Manniesing R., Viergever M., Niessen W.
and gave it a try. I was able to successfully compile the code and run 
the provided test.
The code is easy to read and adopt to the user needs. Many thanks for 
such a nice filter.
I have couple questions though concerning two parts: Multi-Scale Space 
and Diffusion.

1) Diffusion part.
For the stability reason, the TimeStep parameter should be kept 
"reasonably small"
according to the pixel size and scale chosen. However when I try to 
decrease
the TimeStep parameter, the resulting image becomes very close to the 
original one.
So there is no diffusion anymore at small steps or it becomes 
insensitive, right?

2) Multi-scale space part.
In the code example, the scale range was chosen [0.3; 2.0].
The voxel size in the image provided is 0.3125x0.3125x0.5 mm^3.
So is it in a subvoxel range that the user should pick the scale?

Well, in an attempt to merge the low-scale objects, I went to extreme
and set up the range "scales  50 60 70 80 90 100 ".
The program could still happily suppress the noise  and  extract  the 
vessels
quite  similar  to  what  could be  done  with  the scale range  "scales 
0.3 0.482 0.775 1.245 2" .
However this time the lowest scale "50" is even larger than "bounding 
box" (22.8x45 mm^2).

Is the scale part working or I am doing something wrong?

Thanks for your help and support,

Alex