Separation of clumped spots

Hi,

I’m working with an assay that is based on spot detection in cells. I have some images in which I have problems with separating the spots. Since I need the number of spots per cell this gives me huge problems.
I’ve tried to solve the problem with using EnhanceOrSupressFeatures but I still get many of the spots clumped together in my IdentifyPrimaryObject module. I estimate the size of a true spot to 2-5 pixels, but many of them are clumped together to big strings of spots.

I guess that I need some kind of deconvolution to find the local maximum intensity in the spots and separate them.
Is there something I can do to separate the spots better?

I’ll attach one sample image and pipeline.
Thanks!
Karin
DefaultOUT_portugal.cp (9.1 KB)

Hi Karin,

I see two issues at work here, one with the image, the other with the pipeline:

  • The red channel of your image seems somewhat saturated, i.e., regions of the image are hitting the maximum possible intensity limit. This will affect the detection of the spots since there is less contrast to delineate neighboring spots. I would suggest using a lower exposure for this channel if possible.
  • In the pipeline, the settings in IdentifyPrimayrObjects need some adjustment. Try changing the following:
    [list]*]Change the thresholding method to Otsu with 3-class thresholding and the middle class set to Background.
  • Change the lower diameter threshold from 1 to 3. This is a workaround for a bug in the program that will be fixed for the next release.

/*:m][/list:u]
Hope this helps!
-Mark

Hi,

Thanks for the help!
I have improved my pipeline and is satisfied with the segregation but have now run in to another problem. This pops up after maby every fifth image:

Traceback (most recent call last):
File “cellprofiler\pipeline.pyc”, line 309, in run
File “cellprofiler\modules\identifyprimaryobjects.pyc”, line 722, in run
File “cellprofiler\modules\identifyprimaryobjects.pyc”, line 1118, in filter_on_border
File “scipy\sparse\coo.pyc”, line 177, in init
File “scipy\sparse\coo.pyc”, line 213, in _check
ValueError: negative row index found

What should I do to solve this?
Thanks!

Karin

Hi Karin,

Would you mind posting your revised pipeline as well as an image where this occurs?
-Mark

Hi,

Here it comes!

Karin


Pipeline to forum.cp (9.06 KB)

Hi,

I discovered that when I change the lower threshold in the second IdentifyPrimaryObject module from 0.01 to 0.03 the pipeline works. It works for all number over 0.03 as well, but not for 0, 0.01 or 0.02.

Karin

The error is a bug that has been fixed in our code but not yet released. Is the change in the lower bound sufficient for your purposes?
-Mark