Improve yeast colony counting

hi,
I’m trying to convince my boss that using the cell profiler program is as good as using my hours in manual colonies counting!
this is the the pipeline I created using the red colonies yeast cp.
I’m starting with the original picture that I crop using photoshop, making 7.tif
then I run this picture using the pipeline.

the thing is that using the program the colonies number is quite bigger that the result i get whit manual counter. I guess it’s due to the program counting many colonies at the border where there are no colonies.
I tried to crop the picture trying to avoid the light shadows and the glares, but I can’t cut in a way that there are no colonies at the borders.

can you help me with that please!!

thank you




yeast colony counter.cp (9.63 KB)

It appears that the image created by CorrectIlluminationCalculate, when used to correct the image, is introducing aberrations which are then picked up as colonies.

I would insert a couple of modules prior to your CorrectIlluminationCalculate module:

  • IdentifyPrimaryObjects to identify the dish: Since you cropped the image, this is straightforward. Uncheck the “Discard objects outside diameter range…” and “Discard objects touching…” settings, and set “method to distinguish clumped objects” to “None”. Basically, you want the entiure dish to indeitfied as one objects.
  • MaskImage: Set the input image to Orig, “Objects or image as mask” to “Objects” and select the dish object you created above. This better works with illumination correction than what you had before, because the area outside the cropped region is ignored.

In CorrectilluminationCalculate and CorrectilluminationApply, select this masked image as the input image to be corrected. Also, I think you want to reduce the block size in CorrectilluminationCalculate to say, 5.

Then add the following after the IdentifyPrimaryObjects module:

  • ExpandOrShrinkObjects: Set the input object to the dish, and the operation as “Shrink by a specified number of pixels”. Set the number of pixels to 1 or 2.
  • MaskObjects: Set Colonies as the input object, the shrunken dish as the masking object and “Handling of objects…” set to “Remove”. This will remove the colonies which are touching the cropped edge of the dish.

Regards,
-Mark

thank you for the help Mark,

actually at the end I’m just using the pipeline for counting the yeast red colonies. I crop the picture of 4 plates to have one plate at the time (fig.1 attached) then I run the pipeline using the mask attached in the red colony pipeline. so in this case there are no problems at the border but the counting is imprecise when the colonies are too close to each other.



In the colony identification step, I would make the following changes:

  • Method to distinguish clumped objects: Intensity
  • Method to draw dividing lines:Intensity
  • Smoothing filter size: 3
  • Maxima suppression distance: 3

You may have to play around with the latter two settings to optimize the colony identification, but I don’t think you’ll need to adjust it much beyond these values.

Regards,
-Mark

Dear Mark,

using 4 and 4 values and the adjustments you suggested I improved a lot the results!
there are still few imperfections but are minors! thank you so much!!


Good to hear!
-Mark