How to: identify metaphase cells from brightfield images

Hi all,
As I’m presently getting to grips with the execution of a pipeline to identify metaphase cells from brightfield images (I’m aware this is a somewhat difficult task).

I would greatly appreciate if anyone has attempted this or would like to make any suggestions on the pipeline construction.

The metaphases are the rounded cells with the halo’s, ideally the ones where the separating chromosomes are visible see e.g. .

And here are some images which could be used to run test this out.
e.g. G144-G144D7_D2_1_2010y04m22d_15h30m.tif

I’ve also included what I consider to be positives… these are identifiable by the middle stripe.

I’ve my own ideas and a pipeline I’m working on, but before I post that I’m interested to see what others might come up with.

Many thanks in advance,
Amos (4.17 MB)

You may have some luck with using EnhanceOrSuppressFeatures. If you select “Enhance” as the operation, and “Dark holes” as the feature type, you’ll get additional settings allowing you to enhance “dark hole” features, which are dark circular areas surrounded by bright rings.

A quick check of your images indicates that a range of 5 to 25 pixels may work well. The result is an enhanced image in which the dark holes appear bright and everything else is dimmer; this image can then be passed into IdenitfyPrimaryObjs as input.


1 Like

Thanks Mark, that was very useful.