# Segmentation issue - "invisible" labels

**URL:** https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460
**Category:** Support
**Created:** [February 21, 2024, 10:15pm UTC](https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460 "2024-02-21T22:15:51Z")
**Posts on this page:** 4
**Page:** 1

<div class="post-metadata">

### Author: ![Dhartez](https://avatars.discourse-cdn.com/v4/letter/d/9de0a6/32.png) [@Dhartez](https://discourse.slicer.org/u/Dhartez)
#### Post date: [February 21, 2024, 10:15pm UTC](https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460/1 "2024-02-21T22:15:51Z")

</div>

Hi all,

sorry for semi-reposting (I initially thought this was a pyradiomics issue and therefore posted there, but now figured that it is a segmentation issue):

I am having a problem with importing CNN generated segmentations of two brain structures (bilaterally, so 4 segmentations / VOIs in total) on MRI.

When I load the MRI volume and the segmentation file, it looks good, only the light blue VOI isn’t fully recognized as a segmentation/label it seems:

 ![Screen1](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/9/7/972b96b2bd1ed0f59767d6f2e55b5b125f66e42f.jpeg)

But when I add it in the Segmentations module it looks fine:

 ![Screen2](https://us1.discourse-cdn.com/flex002/uploads/slicer/original/3X/2/4/240df37792f5e92a8eb0f0d31cfb825eeadc2ca3.jpeg)

However, out of the four segmentations, I can only edit a single one (the brown one, seg 4), although all four are visible - any segmentation tool that I try on the other three will result in nothing - as if they were not there at all.

How can that be?

The volume and segmentation file are in the folder below (all deidentified of course):

[Link](https://cloudius.meduniwien.ac.at/index.php/s/tRFmMXwUbacZE05)

Many thanks!

---

<div class="post-metadata">

### Author: ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)
#### Post date: [February 22, 2024, 4:49am UTC](https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460/2 "2024-02-22T04:49:52Z")

</div>

You can try upgrading to the current Slicer version (5.6.1) and make sure you save the segmentation into an image with integer type (unsigned char, short, etc).

---

<div class="post-metadata">

### Author: ![Dhartez](https://avatars.discourse-cdn.com/v4/letter/d/9de0a6/32.png) [@Dhartez](https://discourse.slicer.org/u/Dhartez)
#### Post date: [February 22, 2024, 2:40pm UTC](https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460/3 "2024-02-22T14:40:02Z")

</div>

Thanks - I tried the new version, but no change, still not editable.

In terms of saving, what option would that be (i.e. file extension)?

---

<div class="post-metadata">

### Author: ![lassoan](https://sea2.discourse-cdn.com/flex002/user_avatar/discourse.slicer.org/lassoan/32/13_2.png) [@lassoan](https://discourse.slicer.org/u/lassoan)
#### Post date: [February 22, 2024, 2:44pm UTC](https://discourse.slicer.org/t/segmentation-issue-invisible-labels/34460/4 "2024-02-22T14:44:59Z")

</div>

To fix the loading issue, you can convert your numpy array to uint8 before writing to file. Like this:

> <https://github.com/lassoan/SlicerMONAIAuto3DSeg/blob/72abd51b670bf989f359632ab1e9e2841ed268ca/MONAIAuto3DSeg/Scripts/auto3dseg_segresnet_inference.py#L209>
