[Bug] Segmentation Rendering Broken with Gaps in MPR

[adriangaliana]

Describe the Bug

In the release or beta version of OHIF, for some series (not all), when a segmentation is performed or loaded from PACS in the acquisition plane, it displays correctly. However, when switching to MPR view, the segmentation appears broken or corrupted.

Steps to Reproduce

1. Go to https://viewer.ohif.org/ or https://viewer-dev.ohif.org/
2. Open the patient PCAMPMRI-00012 in either basic or segmentation mode
3. Load the T2 segmentation or create a new manual segmentation
4. Switch to MPR view
5. The segmentation appears distorted and displays visible gaps

The current behavior

Stable 3.10.1

3.11.0 Beta.42

The expected behavior

The segmentation should appear continuous and intact in MPR view, without any gaps.

OS

Windows 11

Node version

20

Browser

Chrome v136.0.7103.114

[pieper]

Since there are gaps in the segmentation as seen in the sagittal and coronal views, what you see in the axial view looks correct to me.

Note that sampling binary labelmaps through a transform leads some fundamental and maybe unexpected rendering conditions: https://discourse.slicer.org/t/segmenting-in-a-rotated-volume/29147

[adriangaliana]

Thanks for the information! However, I don't think that's the issue in this case. It worked fine in earlier versions, like 3.8.

It seems there's a problem with the MPR-generated volume. As you can see, the segmentation appears correctly in the Axial view initially, but after performing MPR, it no longer aligns properly in the Axial view.

I also tried changing the acquisition plane and using different segmentations, but none of those options resolved the issue:

https://github.com/user-attachments/assets/0b2cef71-2045-4e78-945f-3e90be8ce963

[sedghi]

Yeah, I've seen this. Honestly, I'm not really sure what's going on. Something seems off, but only with this data. It looks like the data was acquired at an oblique angle.

[sedghi]

I am aware of this, it is a weird bug