Identifying individual wells on a Chip


I am a PhD student in Stem Cell Biology at King’s College London and I am planning to use CellProfiler for image analysis.

The images that I am planning to analyse are of a Chip, with several individually different wells on there (1056 to be precise). Every image displays at least one well. However, as the wells do not cover the whole image region there can be more wells (or parts of wells) on an image than just one.

Now what I plan to do is to identify individual whole wells in an image, crop images accordingly and then analyse the features (cell content, fluorescent marker expression, etc.) inside of every well. Theoretically, I think this should be a relatively ‘simple’ pipeline.

However, there are a couple of issues with this:

  1. The frame of the image shifts a bit as I reach further parts of the Chip; so ‘whole wells’ of interest are not always in the same position.
  2. The surface of the individual wells differs; they contain different topographic features.

Therefore, I was wondering whether it would be possible to identify individual wells on these chips and, subsequently, to (batch) crop the images so that I can analyse them as separate units.

Would this be something that I could do with the DefineGrid option? Or is there a better/more convenient way to do this?

I looked at a previous post (Grid Problem) from someone with a similar problem and I think I could use a similar approach. However, the thing is I’m not using tiles/stitched images at the moment (I prefer to not do this because I think this would require too much memory) and I’m dealing with these ‘shifting images’.

Attached you can find an example image with one of the wells I’m keen to identifying in the top left.

Any help/suggestions are welcome.

Kind regards,