which channels should be used to segment nuclei and whole cell

[EdwardZhao1991]

Thanks for sharing the great work.

I'd like to segment both nuclei and whole cell.

Could you tell me which channels should I use to segment them?

For example, there are multiple DNA channels, how should I select or merge them?

Any comments would be highly appreciated.

[clarenceyapp]

Hi @EdwardZhao1991 ,

sorry for the delay in responding. UnMICST is a nuclei segmenter only. You have to look through the channels and identify one suitable nuclear marker channel. If you want all nuclei, then you should choose the first cycle as long as it is not saturated or blurry. Since cells will likely fall off over cycles, you can also choose one of the last cycles if you only want to analyze cells that have stayed in place. This is why you should not always merge them because, if cells move, then you will be counting cells multiple times.

Once you have the output nuclei probability map files from UnMICST, you can run S3segmenter using MCMICRO to convert them to nuclei masks and segment the cytoplasm and whole cells. You would need to choose a suitable cytoplasm marker and define that as --CytoMaskChan and set --segmentCytoplasm segmentCytoplasm to activate cytoplasm segmentation. By default, S3segmenter will dilate the nuclear masks by around 3 pixels to approximate the cytoplasm. If you have chose a nuclei channel other than from the first cycle, then you need to also specify this as --probMapChan in S3segmenter.

Hope that helps.

[ArtemSokolov]

Hi @EdwardZhao1991,

I hope Clarence's information was helpful. I am going to close this issue due to inactivity. However, feel free to reopen if you have more questions or would like to follow up.