Problems related to identifying objects

I have some pictures of human T cells (SupT1 cell) and I want to compare the intensity of fluoescence of cells without virus or with virus.
I have RGB images. So, I changed my images from color to gray.
I tried to identify cells in the bright field images and then measure the intensity of fluorescence of cells in the fluorescent images.
But, I have some problems when identifying cells as the primary object in the bright field image.
I’ve tried to change many conditions and here is the module of the best among my trials.
The program identifies some wierd objects and they are not cells and don’t have circle shape.
I don’t understand why the program identifies them as the primary objects.
I have 14 cells and among them, 11 cells are circle shape.
I just added my module and the result of identifying the primary objects.
Can you give me some suggestions for my results?
Oh, the green ones are the objects!
cp8.cp (6.73 KB)


Hi,

Do all the cells fluoresce to some degree even if they are positive or negative? If so, you might have better luck identifying the objects with the fluorescent images over the brightfield ones. CellProfiler is optimized for fluorescent images and can sometimes deal with brightfield images but with some tweaking required; see here under #5 “Can CellProfiler handle my brightfield/phase contrast/DIC image?” for details.

Another approach is using EnhanceOrSupressFeatures with dark hole detection; you can enable this using “Enhance” as the operation, “Dark holes” as the feature type, and specify the feature size. This module serves to enhance dark spots surrounded by bright rings. Since your objects seem to be light objects surrounded by dark rings, you need to use ImageMath to invert the image first before using this module.

Hope this helps!
-Mark

Thanks! Your second suggestions worked well. Actually I used “speckles” not “dark holes”.
I have another problem when identifying objects manually.
I just want to figure the unidentified objects out when using automatic methods.
I don’t understand why there are error messages when identifying objects manually and making spreadsheets.
Is there any problem with my pipeline?
cp8-1.cp (5.03 KB)

Without the images, I won’t be able to tell with just the pipeline alone. But could you post the error message that you’re receiving?
-Mark

Oops.
I added the error messages!



Thanks for posting these. This is a bug in this release of CellProfiler (in IdentifyObjectsManually which carries through to ExportToSpreadsheet), but we’ve already noticed it on our end, and will be fixed for our next release.

Regards,
-Mark

Thanks for your kind reply.
So, is there any method to solve this problem except waiting for new release?

Unfortunately, this is no workaround for this issue.
-Mark

Is there aother method to identify some objects whicn cannot be identified by aumatical method?

If your objects are different in some distinguishable way (e.g, the objects you want are round whereas the ones you want to ignore are some other shape), you can use FilterObjects. To do this, use one or more of the measurement modules on the objects in your image, then use FilterObjects to specify the measurement and cutoff value for that measurement to filter out the ones that you don’t want.

The alternative is re-work your image processing prior to IdentifyPrimaryObjects to better refine detection of the objects that you want. Other than using the ColorToGray module, I see that you’re not doing any further processing beforehand.

Regards,
-Mark

Our latest release, r10415, includes the fix for this issue.
-Mark