3D colliding fronts segmentation in branches

[Lisa Rutten]

Hey,

How does the 3D colliding fronts segmentation work in branches? I'm segmenting a bifurcation using one start seed and two end seeds. When using two seeds, the direction of two vector fields form the initial set function and side branches are excluded because the direction of the fields is the same. Does this mean that using three seeds results in three vector fields which form the initial set function and that this function is positive if the direction of all three fields is the same? Or are two initial set functions combined?

In my case, I'm trying to segment one shorter and one longer branch. Can I use three seed points or is it better to perform two segmentations: 1 main artery to branch 1, 2 main artery to branch 2 and combine these segmentations? (if this is possible?)

Thanks in advance!

[Gabriel Maher]

Hi Lisa,

The colliding fronts algorithm will compute separate vector fields for each start/end seed combination and then union the resulting vessel models.

In your case one start seed with two end seeds should work.

Best,

Gabriel

[Lisa Rutten]

Hi,

Thanks for the explanation! While working on the colliding fronts method I discovered that I need different upper and lower thresholds for different parts of the vessel due to image quality, calcifications and stents. Is it therefore possible to segment these parts separately and combine them in one segmentation?

Furthermore, I was wondering if I could adjust the 3D view in such a way that my segementation is shown in the axial, sagittal and coronal plane instead of "behind" these planes.

3Dview_simvascular.PNG (107.85 KiB) Viewed 644 times

Thanks in advance!

Lisa

[Gabriel Maher]

Yes, if you create individual 3D segmentations for each vessel, you can then merge them together in the modeling part of SimVascular.

It is a bit unclear what is going on with the segmentations in your picture. Are you saying that they do not appear in the axial/sagittal/coronal planes? Or are they simply being positioned incorrectly?

[Lisa Rutten]

Hi,

I think the segmentations are positioned incorrectly. Looking at the display, the segmentation is only visible in the 3D view.

Display_simvascular.PNG (124.33 KiB) Viewed 630 times

In the 3D view, it looks like the segmentation is positioned behind the axial, sagittal and coronal planes in the 3D view. With "reinit" I am able to analyze the segmentation, but I cannot compare this to the CT image, like in the article "A Re-Engineered Software Interface and Workflow for the Open-Source SimVascular Cardiovascular Modeling Package".

Source_simvascular.PNG (508.84 KiB) Viewed 630 times

Is it possible to have the segmentation in the CT image so that it is easier to note errors in the segmentation?

Thanks in advance!

Lisa

[David Parker]

Hi Lisa,

There seems to be a scaling problem with the colliding fronts code, assumes that the image units are cm I think. Please open an Issue on this at https://github.com/SimVascular/SimVascular/issues.

Cheers,
Dave