Redo segmentations #7
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
There was a problem hiding this comment.
Woahhh that IC hurts haha. Maybe would require some changes, one I can recommend is definitely changing from background to regular method and go from there. Very odd!
Also GSDMC is definitely better for segmentation only because there is clear contrast, those other channels are very messy... interesting stuff!
There was a problem hiding this comment.
Looks to be empty unfortunately :(((((
There was a problem hiding this comment.
What is the red line? Mean of the diameter?
Also, very interesting how small the nuclei are, but I am not as familiar with the data so it might be expected! Might need some coSMicQC if those diameters are cluster nuclei/over-segmented haha
There was a problem hiding this comment.
45 diameter looks to be pretty optimal! Unfortunately I can't see all of the options though.
One main thing I noticed (that we chatted about in person) was that these nuclei look very messy and look to contain mycoplasma, in my opinion. I have only seen it once and these nuclei look very similar to what I saw for mycoplasma.
As we discussed, it would be good to ask the collaborators to test the cells to ensure no contamination.
There was a problem hiding this comment.
I don't see any examples in this file for what the cell segmentations look like, so I can't provide feedback on that unfortunately :(((
| # save to a dict for later use | ||
| for img in nuclei: | ||
| img = normalize(img) | ||
| masks, flows, styles, diams = model.eval(img, channels=channels, diameter=50) |
There was a problem hiding this comment.
Totally agree with this diameter!
| #!/usr/bin/env python | ||
| # coding: utf-8 | ||
|
|
||
| # This notebook focuses on trying to find a way to segment cells within organoids properly. |
There was a problem hiding this comment.
| # This notebook focuses on trying to find a way to segment cells within organoids properly. | |
| # This notebook focuses on trying to find a way to segment cells properly. |
Same comment as in the nuclei segmentation script.
|
|
||
| # ## Cellpose | ||
|
|
||
| # In[ ]: |
| # get masks for all the images | ||
| # save to a dict for later use | ||
| for img in imgs: | ||
| # masks, flows, styles, diams = model.eval(img, diameter=diameter, channels=channels) |
There was a problem hiding this comment.
Recommend removing commented out code.
| # This notebook focuses on trying to find a way to segment cells within organoids properly. | ||
| # The end goals is to segment cell and extract morphology features from cellprofiler. | ||
| # These masks must be imported into cellprofiler to extract features. |
There was a problem hiding this comment.
Consider updating this to reflect the cytoplasm issue checking.
This PR performs segmentations within python to import masks into CellProfiler later for feature extraction