Help identifying individual bacteria in dense colonies

I’ve been experiment with cell profiler for a while trying to get a good pipeline assembled for identifying bacterial cells.

I’ve tried a large number of permutations of algorithms and parameters for IdentifyPrimaryObjects (and greatly appreciate the exhaustive documentation), but I seem to be unable to avoid the problem of large clusters of cells being treated as a single cell.

I’ve added as an attachment a sample image file, a sample output file, and an image of the pipeline I’m using.

Any help would be greatly appreciated! Thanks!







Hi,

I’m attaching a pipeline which should get you closer, although it’s not perfect. I had to fine-tune the declumping settings in test mode to find a “sweet-spot” of sorts.

My suspicion that a higher resolution image or a longer exposure time might help with optimizing the declumping task. The graininess of the image tends to cause spurious breaks in the objects, so my suspicion that a higher resolution image or a longer exposure time might help with increasing the signal-to-noise ratio and/or increasing the effective spacing between the cells.

Regards,
-Mark
2011_03_16.cp (4.3 KB)