Simple analysis of very tricky IRM images


I have searched for this topic but found nothing, IRM is not such a popular technique anymore.

I am counting cells (and measuring area) that attach to glass slides, the imaging technique is interference reflectance microscopy, so the images are dark, have interference patterns on them and the cell outlines are very similar to background levels, and they can be lighter or darker than the background depending on the protein the glass is plated with…

Please see the images for an idea of how it looks;

The darker (picture A) and lighter (picture B) blobs are the bottoms of cells that have attached to the glass. I want to have a pipeline that counts them, and measures surface area per cell attached.

Is this beyond the capabilities of Cell profiler? It feels like the counting is relatively simple for a person, the shapes are distinct, but because of the interference and the minimal differences in signal to background, that it might just be impossible to use Cell Profiler for this.

I have tried everything I can think of in Cell Profiler, but I am aware that I am a total beginner with the software and thought maybe there might be some kind of image processing that could help here.

Any ideas? Any tips at all on what to try would be greatly appreciated!


We think it’s rather difficult to use CellProfiler for this. Since both the background and foreground (cells) are inconsistent, you have to identify fringes in the background, then masks them out, then build 2 pipeline for bright cell cases and dark cell cases. All that doesn’t sound worth it.

I suggest you to have a look at ilastik and try

Good luck.

Thanks! I’ll check it out!