Viewers: handle fractional SEG objects

[pieper]

Request

Related to #1345, the segmentation is encoded as fractional but we currently don't load it.

Instead we throw this as an error:

Uncaught (in promise) Error: Fractional segmentations are not supported

What feature or change would you like to see made?

We should probably just threshold this to binary for now. Better of course would be to handle the fractional case as a kind of heatmap overlay.

Why should we prioritize this feature?
It's part of the DICOM standard and we have examples of data encoded this way.

[JamesAPetts]

Better of course would be to handle the fractional case as a kind of heatmap overlay.

Yeah this one is interesting, this would be the best thing to do, but it'd essentially be a different data type in cornerstoneTools. I think the simplest thing to do would be to save each segment as its own labelmap as per standard and give it its own colorLUT that is the same color with different alpha per value?

Is it common to have multiple fractional segments in one DICOM SEG on one referenced series?

[pieper]

Is it common to have multiple fractional segments in one DICOM SEG on one referenced series?

Not sure if there's any consensus on this, but if it's okay according to the standard then probably somebody will do it. It would be great if we could work on promoting best practices for how people should structure their SEG instances. (Comments @fedorov ?)

[JamesAPetts]

I'm sure its DICOM compliant, but the web dev in me is already screaming internally about the potential consequences on resource allocation :).

[fedorov]

From a quick look, I am not sure how this is related to #1345 - can you clarify? Also, considering we don't have any fractional SEGs in TCGA, and it's not in IDC MVP, I don't consider this to be IDC priority.

I have very limited experience with fractional SEG, I don't think I have seen any examples other than the objects produced by BrainLab, and I don't know of any tools that create fractional SEGs.

[fedorov]

We confirmed with @pieper that the segmentations in that kidney collections are binary, so the question is why that fractional error is coming up.