I’m posting a pipeline that seems to work on this image pair; see the module notes for more details… However, it may require modification as you input your other images. Also, for future reference, we recommend that you use loss-less image formats such as TIF or PNG for processing rather than JPGs which are small but lose information.
Some caveats apply here (which I also mention in the module notes in the pipeline):
- This pipeline reports the measurements associated with the single object with the largest intensity measurement in the image. If the fungus is smaller than the other objects found, this approach may not work (but is still more likely to work than an area-based measure, I think).
- To truly make a good comparison between images, you must ensure that all images are acquired at the same resolution, magnification, lighting conditions, etc.
- For the color subtraction approach to work, the relative color balance mus be maintained. If possible, do not change the color of the background material that the plate is resting on.
Based on other user requests for fungal colony counting (such as this one which was fairly difficult), it seems like you have an advantage in that you’ve chosen a background color that contrasts well against the fungus; it’s something which I was able to exploit in the pipeline. Was that color deliberately chosen for that reason? If so, it may be something that I recommend to others in the future, if they have that option.
2011_04_25.cp (10.9 KB)