Problem loading objects with LoadImage

I’m not sure if this is a bug or whether I have done something wrong. I am trying to mask clumps of cells from a series of time lapse images of proliferating cells. I have split this process up into a few separate pipelines. My steps are:

  1. Run a “mask explore” module that does illumination correction and segmentation to find the clumps. Object identification is loose, to include many false positives. The outputs are:
  • the illumination corrected image
  • the “clump” objects
  • data and profile for CellProfiler Analyst
  1. Explore the objects in CPA to come up with a set of rules that can eliminate the false positives

  2. Run a “object to mask” module in CP that should:

  • load the illumination corrected image
  • load the 1st generation “clump” objects
  • measure multiple parameters for the “clump” objects on the image
  • use the FilterObjects module and the CPA rule set to eliminate undesired objects
  • use the filtered clumps to create and save mask files for use in later steps in subsequent pipelines

I seem to be having a problem loading the clump objects in the first step of the 2nd CP pipeline. There is an error alert with every cycle. Strangely, this error only appears when I run the “Analyze images…” but when I run the same pipeline in F5/step-by-step mode I get no error and the output is as expected.

Error message detail:
Traceback (most recent call last):
File “/Applications/”, line 309, in run
File “/Applications/”, line 2120, in run
File “/Applications/”, line 258, in add_image_measurement
File “/Applications/”, line 324, in add_measurement
KeyError: ObjectsFileName_IMAGE_FOR_clumps

I am using the latest stable release (r11710), running OS X Lion 10.7.4. I have attached the 2 pipelines, the CPA rules and a set of example input and output images.

Ben (1.29 KB)

5-4_clump_to_mask.cp (6.6 KB)
5-4_mask_explore.cp (15.2 KB)

Hi Ben,

The error message you are receiving is one that we have fixed in our code, but not yet released. As a workaround, you can re-identify the objects from the original image within the 2nd pipeline.


Thanks for the explanation. The workaround you suggest is what I’ve done so far.