# Vessel Segmentation for carotid artery with stenosis

**URL:** <https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272>\
**Category:** Support\
**Tags:** segmentation, vessels\
**Created:** [January 16, 2023, 7:02pm UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272 "2023-01-16T19:02:20Z")\
**Posts on this page:** 10\
**Page:** 1

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 16, 2023, 7:02pm UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/1 "2023-01-16T19:02:20Z")

</div>

Hi all,

I am pretty new to 3D slicer, and would very much appreciate your help 🙂

I am trying to segment and export a .stl file of the carotid artery with stenosis based on CT images with contrast.  
Since I am planning on handling 100+ patients’ data, I am trying my best to automate the process as much as possible ( and minimize user variability).

Granted, I have tried flood-filling / grow from seeds, and scissor-ing from the 3D view (threshold 235-600 HU). For all 3 methods, I need to manually erase/paint the lumen for each image due to the plaque and stenosis. Due to the low resolution when zoomed-in and the small cross-sectional area at the region of stenosis, this erase/paint process always leads to high variability.

Is there a more sophisticated way to achieve this? Or is there a guideline that I could follow?

Any guidance or information would be deeply appreciated.

Thank you!

---

<div class="post-metadata">

**Author:** ![pieper](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/pieper/32/8_2.png) [@pieper](https://discourse.slicer.org/u/pieper)\
**Post date:** [January 16, 2023, 8:03pm UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/2 "2023-01-16T20:03:16Z")

</div>

Yes, it’s a hard problem. In terms of improving the consistency for small vessels, you can try increasing the segmentation resolution (use the segmentation geometry button) so that the labelmap has more pixels to define the shape. Also it’s likely that a machine learning algorithm like those in MONAI Label could be trained to do this task, although I haven’t seen it done yet.

---

<div class="post-metadata">

**Author:** ![chir.set](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/chir.set/32/66982_2.png) [@chir.set](https://discourse.slicer.org/u/chir.set)\
**Post date:** [January 16, 2023, 10:59pm UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/3 "2023-01-16T22:59:07Z")

</div>

Can you post your worst case anonymized dataset?

---

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 16, 2023, 11:02pm UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/4 "2023-01-16T23:02:11Z")

</div>

![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/6/f/6f3c977b7bc55118d37fc55b78a641e6b594b20f.png)  
 ![image](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/0/d/0de569268e8dbedbfe89e9aefd1188b06d535799.png)

Here is an example of the carotid artery at the site of stenosis. The white blob is the plaque.  
This is a result of scissoring, and as you can see it reads the plaque as part of the lumen, which isn’t something I want. Moreover, it is difficult to discern where the limits of the lumen/vessel are at.

---

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 17, 2023, 6:05am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/5 "2023-01-17T06:05:01Z")

</div>

Thanks for the reply.

1. The segmentation geometry was insufficient to overcome this problem.

2. I am looking into MONAI label. Would it run with Python 3.10.9 on Windows 10???

---

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 17, 2023, 7:23am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/6 "2023-01-17T07:23:15Z")

</div>

Can I use flood filling in MONAI label module? Or does it have to be grow from seeds?

---

<div class="post-metadata">

**Author:** ![chir.set](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/chir.set/32/66982_2.png) [@chir.set](https://discourse.slicer.org/u/chir.set)\
**Post date:** [January 17, 2023, 7:27am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/7 "2023-01-17T07:27:48Z")

</div>

> [@hoochupapa](#):
>
> an example of the carotid

I meant, a worst case sample volume as NRRD or anonymized DICOM, through a hosting link.

---

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 17, 2023, 7:30am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/8 "2023-01-17T07:30:21Z")

</div>

Oh I see. I can’t share that due to my company’s IP policy ☹

---

<div class="post-metadata">

**Author:** ![gaoyi.cn](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/gaoyi.cn/32/1422_2.png) [@gaoyi.cn](https://discourse.slicer.org/u/gaoyi.cn)\
**Post date:** [January 17, 2023, 7:34am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/9 "2023-01-17T07:34:26Z")

</div>

what OS are you using? I can build an extension for vessel segmentation for you so you can try on your data.

---

<div class="post-metadata">

**Author:** ![hoochupapa](https://avatars.discourse-cdn.com/v4/letter/h/9dc877/32.png) [@hoochupapa](https://discourse.slicer.org/u/hoochupapa)\
**Post date:** [January 17, 2023, 7:35am UTC](https://discourse.slicer.org/t/vessel-segmentation-for-carotid-artery-with-stenosis/27272/10 "2023-01-17T07:35:38Z")

</div>

That sounds wonderful! I am running on Windows 10. I have plenty of RAM and GPU, thankfully.