Issue with cellID between nuclear and cytoplasmic masks

[josenimo]

Dear MCMICRO/mcquant developer(s),

I am trying to analyze HeLa cells from cells culture as a proof of concept image analysis. I am having trouble after segmentation. I successfully segmented both nuclei and cytoplasms for most cells, however I can't think of a simple way to match every nuclei to their cytoplasm mask (CellID), and HeLas sometimes have more than a single nucleus..

I would like to cluster cells into different groups, based on the intensities of each marker inside the nucleus or nuclei, and in the cytoplasm for each cell. The less elegant solution I am thinking about is to just measure the entire cell mask (including nucleus) but then interesting patterns will be diluted at different rates depending on the cell size.

Have you run into this issue?

Thanks in advance for the help Best, Jose

[ArtemSokolov]

Hi @josenimo,

If the masks were generated by S3segmenter, CellIDs should match between all mask files. Are you seeing a discrepancy?

[josenimo]

Hey @ArtemSokolov,

Hmm, I think I might have a problem understanding the output from mcquant. I performed MCMICRO using unmicst/s3segmenter using the cytoplasmic segmentation option. mcquant produced two csv tables. _cell.csv and _cellRing.csv. I assumed, now not so sure, that since the areas are much larger in the _cell.csv table, that these were the cytoplasmic areas (for the same CellID). and that the ones from _cellRing.csv are from the nuclei mask, also since their DNA1 intensities are much larger as well.

Could you go into a little detail of what exactly mcquant is measuring, (I am familiar with the scikit-image region properties) and what is the nomenclature for the output?

Thanks, and sorry for the confusion, there is not much in the github README.

Best, Jose

[DenisSch]

Hi Jose,

• "cell" refers to the mask using membrane information
• "cellRing" refers to the mask simply extending the nuclei information with a fixed distance