Need Help detecting how Blue a colony is

Hi I have been making progress with cell profiler but am having a few issues and was wondering if anyone might be able to help. I am trying to take a plate and to detect all of the colonies that are growing on a plate, then determine how blue each of those colonies is. I am able to detect the blue colonies alone and all of the colonies, but I have not been able to detect how blue each of the colonies is. One other issue that I have been facing is how to get the grid coordinates for each of the colonies. I have attached my results so far, along with my current pipeline. Any help would be greatly appreciated. Thank you
PlatePipeLineExample.cp (16.1 KB)

Could you upload an example image as well (with the colonies plus the template)?

Hi Mark, sorry about that, I have attached the sample image and the template that I used to generate the results that I posted earlier. Thank you

Hey Alos,

Sorry I haven’t been able to help much with this. I’m working on a side collaboration now and it’s been leaving me very little time to tinker with CP/CPA.

Anyway, I think you’re going about quantifying “blueness” the wrong way by trying to identify each colony as an object via IdPrimaryObjects and then measure it. I’m ont a CP expert, but there should be a way to use the grid as your segmentation and then quantify the contents of each cell.

Dave/Kate/Mark… any idea how to do this or something similar?

So you have divided the RGB into each color.
You identify all the wells.

But you can’t find how blue is each well?

Am I undertanding the issue properly?

If so, simply use the meassure intensity with all the wells as objects doing the meassures in blue image?

I haven’t used microtiter plates with CP, but you can see the XY coordinates in exported spreadsheet/database.

On the spreadsheet you can probably stablish rules such as if x=3 ;pos=1 if x=6 pos=2. if y=2; pos=A ; if y=5 pos=B…

Thanks Adam, thanks Feel I will give that a shot. Cheers

Hi Alos,

If you give this a try, don’t be surprised if you find it produces an error like this:

ValueError: shape mismatch: objects cannot be broadcast to a single shape

or something similar indicating a mismatch in image sizes.

This is from a bug from size differences between the matrix containing the objects and the images they came from in some modules. We are working on fixing this for the next release.


Thanks Mark

Our newest release, r10415, should fix the size mismatch problem.