Page 1 of 1
3D Segmentation Capability
Posted: Thu Sep 07, 2017 5:53 am
by jmuskat
Hello,
I'm looking to use the 3D segmentation tool to model the Circle of Willis with an aneurysm. I've tried planting three 'inlet' seeds with six 'outlet' seeds.
Is there a way to use multiple "normal/start" seeds for segmentation? Also, what should I be aiming to have the threshold set at? I keep getting partial vessels; concomitantly, I'm seeing only one or two "normal/start" seeds with vessel formation.
Thank you for the help.
Joseph
Re: 3D Segmentation Capability
Posted: Mon Sep 11, 2017 9:56 am
by gmaher
Hi Joseph,
The 3D segmentation tool allows you to place as many start seeds as you like, these are the "inlet" seeds you mention, so you should be able to use multiple of them. For completely separate vessels it does tend to work better to segment them separately however.
The 3D segementation stops as soon as one of the outlet segmentations reaches an inlet, therefore the 3D segmentation works better if you place the seeds further back in the vessels to give all the outlets a chance to merge.
The threshold should be set to capture the minimum and maximum vessel intensity, this prevents the segmentation from leaking out.
If you have any screenshots of the problems you are having those would be useful to help you out more.
Regards,
Gabrie;
Re: 3D Segmentation Capability
Posted: Mon Sep 11, 2017 10:18 am
by jmuskat
Hello Gabriel,
Thank you for the information. That makes more sense.
I can share photos in a little bit, but I wanted to check how the 3D segmentation progresses ("stops at the outlet"). I've placed the seeds at the entrance locations for my inlets, but the tool keeps segmenting down one vessel/never branching off.
Should I put seeds at a bifurcation?
Joseph
Re: 3D Segmentation Capability
Posted: Mon Sep 11, 2017 11:12 am
by jmuskat
- Example.PNG (265.9 KiB) Viewed 701 times
As you can see in the image, I've got two inlets and two outlets leaving an aneurysm in the Circle of Willis. However, one inlet seed/outlet pair doesn't have any segmentation. I've tried various threshold levels, but I can't get it to work.
In addition, The aneurysm itself isn't being filled--as shown by the axial slice.
Joseph
Re: 3D Segmentation Capability
Posted: Mon Sep 11, 2017 3:44 pm
by gmaher
Thanks for uploading the picture, that makes it more clear. I would suggest you try using one starting seed and try to place ending seeds where you would like to produce a segmentation, that might work better.
Re: 3D Segmentation Capability
Posted: Tue Sep 12, 2017 6:13 am
by jmuskat
- Ex.PNG (132.74 KiB) Viewed 686 times
Hi Gabriel,
I haven't had a ton of luck with getting the aneurysm to fill.. It seems that the colliding fronts move towards the ends instead of filling all "neighboring" space. Therefore, I've tried to place an end in the aneurysm instead. First, not sure if this will work?
Secondly, I'm seeing segmentation on all fronts being very spotty. I've tried to figure out the thresholds, and I wanted to know if the numbers in the bottom right of the image (100 | 75) are able to be used to 'preview' the correct thresholds? I tried to segment with these set, but nothing happened as shown.
Joseph
Re: 3D Segmentation Capability
Posted: Wed Sep 13, 2017 8:25 am
by gmaher
Hi Joseph,
Thanks for providing more details about the aneurysm you are trying to segment. In this case it may be faster to create a path through the aneurysm and use the 2D segmentation tools to manually draw the shape of the aneurysm, this should give you more flexibility to get the exact shape you want. Unfortunately at this point I don't know why the 3D segmentation tool is not working for your case.
I have made a note of the issues you have had with the 3D segmentation tool and will try to address them with future bug fixes, so thank you for bringing this up!
Gabriel