[Insight-users] compute an abs. and rel. distance-field image based on a start front, an obstacle image and a finish front

[Dan Mueller]

Hi Roman,

See my response inline below.

Cheers, Dan

On 22 August 2013 20:32, Dr. Roman Grothausmann
<grothausmann.roman at mh-hannover.de> wrote:
> Hello Dan,
>
>
> Many thanks for that hint, I got it working with that. May I suggest these
> additions to be included in ITK. (They even do not seem to be part of Your
> contribution: Fast Marching Minimal Path Extraction
> http://www.insight-journal.org/browse/publication/213)
> These additions would be great and would make the itkFastMarchingImageFilter
> a bit more intuitive since most ITK-filters expect input through images.
>
> Four questions I still have:
> - Is it OK to set the trail points to the same set as the alive points,
> i.e.:
>
>     filter->SetAlivePoints(ANodes);
>     filter->SetTrialPoints(ANodes); //try with trial points = alive points
>
> Or should I actually create an image only consisting of the voxels adjacent
> to the start voxels?
>
> - Should I make sure the speed image is < 1 at the positions of the start
> voxels, i.e. kind of mask the speed image with the start front image?

Not required. The speed image under the start indices can be 1.0.

> - Does the fast-marching distance field consists of linear distances or
> squared distances and does that for some reason depend on the image
> dimension?

The fast marching filter actually returns an arrival function, not a
distance map per se. When the speed image = 1.0, then the arrival
function is the same as distance map with linear distances.

> - What is itkFastMarchingImageFilter::SetBinaryMask meant for?