Count cells in low-contrast brightfield image


I need to count cells in a low-contrast brightfield time-series. I tried the PHANTAST plugin (even though it was designed for phase contrast images) and it worked surprisingly well, but it only output data for the final frame and I need an output stack for the entire time series. Is there a version of the PHANTAST plugin that will output a stack of images? Thank you. I attached a sample frame from the time-series.PCa CD52 Apoptosis 2BottomRight-1.tif (7.0 MB)

Do you need to account for the same sell in multiple Images or would a count for each 2D plane work?

Thank you for replying, Susan! No tracking needed, a 2D count for each frame would suffice. I have tried the PHANTAST plugin and Find Edges, but there are very low contrast cells in some frames that need counting or segmentation. Thank you.


Just did a very quick try using Machine-Learning… Can you provide more info on which cells you really consider real. And some post-processing etc. splitting and filtering would be needed to get better results (but is possible)

Sebi (from ZEISS)

Thank you, Sebi! My client needs the flat, larger cells counted. The small round cells are killer cells. The tough part is the flat, larger cells are cancer cells with very low contrast. I’ve attached a really tough tiled image that I have to process and this point, I cannot come up with anything but brute force manual counting. I welcome your suggestions. C1-PCa CD52 Apoptosis 2TopRightCalibFr52.tif (7.0 MB)


one question: Are those images acquired inside wells close to the well edges? Because the shading is crazy. I have seen this many times when using bbrightfield etc. in normal micorscopes close to the well edges. Can you provide an annoted image with the cells you want to segement? I really have ar hard time to see what you exactly mean.

(Attention: commercial answer): I was one of the responsible persons for the development of tghe CellDiscoverer7, which has a built-in Phase Gradient Contrast Method using an Half-Pupil with an adjustable rotating angle. This would allow to image such cell close to the well edges with high contrast and little or even no shading.
If you ar interested: CellDiscoverer 7

1 Like