S3segmenter not finding all cells (parameters issue?)

[O-kaneko]

Hi, I've been having issues with s3segmenter not finding all cells within my TMA cores.

I've tried with both the original params.yml file and with alterations such as --nucleiRegion localMax --logSigma 1 70 and with specifying --CytoMaskChan [multiple channel numbers]. I found that modifying the params file did increase the amount of cells identified in the cellOutlines.ome and the nucleiOutlines.ome, but immediately post-run, the params.yml file reverted to its original form, removing my changes. Is this supposed to happen?

Attached below are the cell segmentation results with modified params to include local max and cytoplasm (in yellow, DAPI (detection channel) in blue). Is there anything else I can modify to make the segmenter more accurate? Thanks.

[ArtemSokolov]

Hi @O-kaneko,

I found that modifying the params file did increase the amount of cells identified in the cellOutlines.ome and the nucleiOutlines.ome, but immediately post-run, the params.yml file reverted to its original form, removing my changes. Is this supposed to happen?

Can you please share the params.yml that you used? It definitely should not be reverting, so I want to see if I can reproduce your issue.

Unfortunately, I can't help with parameter settings for more accurate segmentation. That's probably a question for S3seg developers. (cc @clarenceyapp, @Yu-AnChen )