# Cortical vessel segmentation

**URL:** <https://discourse.slicer.org/t/cortical-vessel-segmentation/34152>\
**Category:** Support\
**Created:** [February 5, 2024, 1:09pm UTC](https://discourse.slicer.org/t/cortical-vessel-segmentation/34152 "2024-02-05T13:09:00Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![anja\_pantovic](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/anja_pantovic/32/17063_2.png) [@anja\_pantovic](https://discourse.slicer.org/u/anja_pantovic)\
**Post date:** [February 5, 2024, 1:09pm UTC](https://discourse.slicer.org/t/cortical-vessel-segmentation/34152/1 "2024-02-05T13:09:00Z")

</div>

Hello,  
I am trying to segment cortical vessels fro gadolinium enhanced t1 MRI images. I tried using local thresholding as well as vessleness filtering, however, scalp and even white matter interfere with the segmentation.

I don’t know how to get a clean segmentation of vessels and I think the vessels are visible enough in the images, as given below.  
Do you have any suggestions on what could help?

Thanks!  
Anja

 ![Screenshot 2024-02-05 at 14.08.38](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/1/c/1cc78f4c1f16e9f6eac74c0e3c735e8f4f56dd94.jpeg)

---

<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:** [February 5, 2024, 1:46pm UTC](https://discourse.slicer.org/t/cortical-vessel-segmentation/34152/2 "2024-02-05T13:46:04Z")

</div>

That’s a hard problem and I don’t think there’s anything automated. Eventually there will be a deep learning solution I’m sure. If you don’t find anything you could try training something using MONAI Label or nnU-Net.

---

<div class="post-metadata">

**Author:** ![kopachini](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/kopachini/32/65543_2.png) [@kopachini](https://discourse.slicer.org/u/kopachini)\
**Post date:** [February 7, 2024, 5:01am UTC](https://discourse.slicer.org/t/cortical-vessel-segmentation/34152/3 "2024-02-07T05:01:34Z")

</div>

The best approach is manual segmentation, or if you want to speed up, do a treshold and then manual corrections (used with treshold settings also) but it won’t work for small structures.

---

<div class="post-metadata">

**Author:** ![dokay1](https://avatars.discourse-cdn.com/v4/letter/d/b4bc9f/32.png) [@dokay1](https://discourse.slicer.org/u/dokay1)\
**Post date:** [February 7, 2024, 5:37pm UTC](https://discourse.slicer.org/t/cortical-vessel-segmentation/34152/4 "2024-02-07T17:37:48Z")

</div>

If you have good quality T1-pre images (without motion) using the same protocol, doing subtraction of T1post - T1pre can give you the vasculature map and then you can clean that up as needed.