Error when extracting centerline from using vmtkcenterlines

[Moodaeng Theparitt]

Hi there,

      I would like to extract a centerline from blood vessels ( please find the attach file ). However the program runs forever and show  error in vtkMath. 

[Rick Izzo]

Hey Moodaeng, 

Thanks for the email. I checked out your file, and there were a number of issues with it. 

1) the surface had a number of defects in it. There were unconnected regions & degenerate triangle (highlighted in pink), as well as holes in the surface (outlined in blue). I was able to clean these up with an external tool and have attached the result to this email (test-fixed.stl). 

 

2) The surface is very rough (I imagine that this was segmented with a fairly low resolution scan?). I ran it through vmtksurfacesmoothing to get the following output.

At this point I ran the file through the regular vmtkcenterlines script, and was able to generate the following centerlines in about 10 seconds on my machine

This should work for you, but I would caution that the output you will receive right now will not be what you expect. The centerlines algorithm in vmtk does not handle cyclic segments. This is normally not an issue, as there are relatively few regions in the arterial system where vessels form a loop (yes, the circle of willis is a common exception, but it doesn't appear that the patient's left A1 or right P1 segments are intact in this scan anyways). However, in cases like yours where the input image is low resolution, a fast marching or seeds based image initialization method can accidentally initialize phantom vessel connections which are then generated in the surface (you may see this behavior if you modify the advection/propogation parameters of the levelset evolution algorithm as well). Oftentimes this presents as a merging of close lying vessels segments into a nonsensical loop. See the following photo for an example of what I mean in your geometry. 

If you spend time with the colliding front initialization method you may receive better results (though there is a bit more manual effort required). The other option you have is to export your surface as an STL, and use an external CAD editor to manually cleanup and repair sections which are clearly not part of the patient's anatomy, then perform the centerlines extraction on the cleaned geometry. I've used both methods when dealing with this behavior in the past, but if you're not a CAD expert, I would recomend trying to initialize the vessel segments more carefully (with the colliding fronts method) first. 

Hope this helps, let me know if you have any other questions. 

Rick Izzo

Fellow, Western New York Prosperity Program

PhD Candidate, University at Buffalo

875 Ellicott St., Eighth Floor, Buffalo, NY 14203

Rick Izzo

Fellow, Western New York Prosperity Program

PhD Candidate, University at Buffalo

875 Ellicott St., Eighth Floor, Buffalo, NY 14203

Phone: 716-435-3362, Email: rizz...@gmail.com

On Mon, Jan 29, 2018 at 2:50 AM, Moodaeng Theparitt <thep...@gmail.com> wrote:

Hi there,

      I would like to extract a centerline from blood vessels ( please find the attach file ). However the program runs forever and show  error in vtkMath. 

--
You received this message because you are subscribed to the Google Groups "vmtk-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to vmtk-users+unsubscribe@googlegroups.com.
To post to this group, send email to vmtk-...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[theparit peerasathien]

Hi Rick, 

     Thank you very much. Really appreciate your time. Thanks

Sincerely

Moodaeng

--
You received this message because you are subscribed to a topic in the Google Groups "vmtk-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/vmtk-users/jCicOf2LxM0/unsubscribe.
To unsubscribe from this group and all its topics, send an email to vmtk-users+unsubscribe@googlegroups.com.

[theparit peerasathien]

Dear Rick,

        Could you let me know what program that you use for checking open-face polygon ( like the first picture ) please? 

Sincerely

Moo

[Rick Izzo]

Meshmixer. Available at: http://meshmixer.com/

Rick Izzo

Fellow, Western New York Prosperity Program

PhD Candidate, University at Buffalo

875 Ellicott St., Eighth Floor, Buffalo, NY 14203

Phone: 716-435-3362, Email: rizz...@gmail.com