search

hci-unihd / antibodies-analysis-issues

Issue tracker for problems in the antibodies analysis workflow.
0 stars 0 forks source link

Cell Segmentation in H01 - plate5rep3_20200507_113530_429 - H01-0007 #77

Open tischi opened 4 years ago

tischi commented 4 years ago

{"plateName":"plate5rep3_20200507_113530_429","siteName":"H01-0007","pixelLocation":[1464.7272727272727,24055.272727272728,0.0],"analysisVersion":"0e99da3c9fbbdfa66428cb18b2217b93bcd97d51"}

@constantinpape

On this plate cell segmentation in H01 looks surprisingly decent, given that there is hardly any serum signal.

The cell segments are however the smallest of the whole plate.

Anyway, even if it works I am not sure we should rely on it...

What else do you want me to check? Should I check in other plates?

Screenshot

constantinpape commented 4 years ago

While the segmentation does not look to bad, I think we have a couple of issues of "trapped nuclei", this means a segment that only gets assigned its seed. This can really be an issue for the infected cell classification.

constantinpape commented 4 years ago

What else do you want me to check? Should I check in other plates?

No, I think it's fine it will look similar everywhere else (I mean for WellH01).

Could you just say if there was anything weird with the segmentation in this plate overall?

tischi commented 4 years ago

Could you just say if there was anything weird with the segmentation in this plate overall?

nope; nothing obvious