Exemplar-001 nuclei segmentation is identical to the cell segmentation

[LucaMarconato]

It appears that the nuclei segmentation in exemplar-001 is incorrect as it coincides with the cell segmentation (see screenshot from FIJI).

The two files are located in exemplar-001/segmentation/unmicst-exemplar_001/ and have filenames cell.ome.tif and nuclei.ome.tif.

CC @keller-mark, who originally reported the bug in scverse/spatialdata-notebooks#52.

[LucaMarconato]

I see a similar problem also with exemplar-002, in fact the segmentation files in qc/s3seg/unmicst-1/ for cells and nuclei (cellOutlines.ome.tif and nucleiOutlines.ome.tif), show the same segmentation.

[ArtemSokolov]

Thanks, @LucaMarconato. Is this with default parameters?

@clarenceyapp Can you take a look when you have a chance?

[LucaMarconato]

The files have been generated by @melonora; can you please comment on this?

[clarenceyapp]

By default, cytoplasm segmentation is switched off. So only nuclei are segmented. Users have historically wanted a mask called cell to always get written out even if it's drawing intensities from the nucleus so both files (nucleus and cell) will be duplicates. To get a true cell, cytoplasm, and nucleus mask, the segment cytoplasm flag must be on. Clarence

…

________________________________ From: LucaMarconato ***@***.***> Sent: Saturday, January 6, 2024 11:11:18 AM To: labsyspharm/mcmicro ***@***.***> Cc: Yapp, Clarence ***@***.***>; Mention ***@***.***> Subject: Re: [labsyspharm/mcmicro] Exemplar-001 nuclei segmentation is identical to the cell segmentation (Issue #531) The files have been generated by @melonora<https://github.com/melonora>; can you please comment on this? — Reply to this email directly, view it on GitHub<#531 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AFLKVSZTBU7H2ACY4PM5JLLYNFZSNAVCNFSM6AAAAABBPRVQ6OVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNZZG42DCNJSG4>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[LucaMarconato]

Thank you for the clarification @clarenceyapp!