Overlap of objects

Image Analysis
#1

Hi Mark,

I am a new user of Cellprofiler.Thanks for providing such a wonderful software for image analysis.

My work is to analyze image clusters showing the fluorescence variation in time course. The fluorescence signal concentrates in the nuclei of the cells. In my analysis, i have to track the cells, do segmentation, quantify the integrated fluo intensity of each cell and finally draw different curves reflecting the variation of fluo of each individual cell. After i perform the identifyPrimaryObjects module, I have to execute the IdentifyObjectsManually to “delete some objects (cells)”, and the EditObjectsManually to “add some objects” because some cells are nearby the mitosis and adjacent.

So can i overlap these two groups pf objects together for the following fluo measurement? the current version seems to lack such a function.

#2

Hi,

You are correct in that there is no straightforward way to do this, but a workaround could be to use ConvertObjectsToImage for both sets of objects, use ImageMath to add the converted images together, and then use IdentifyPrimaryObjects again on the combined image.

However, before going down this path, it might be worth checking to see if the segmentation/thresholding settings in your initial IdentifyPrimaryObjects could be optimized. Would you mind posting a couple of representative images in which you’ve had to add objects manually? (If you can indicate which objects are missed in your follow-up post for these images, that would be helpful too).

Regards,
-Mark

#3

Here are the figures and my settings. And I met with a error report (the final pic) when I try to use the IdentifyObjectsManually alone. I wonder if it was a bug.

Best

Peng

Slide5.jpg720×540 75.8 KB


Slide4.jpg720×540 79.6 KB


Slide3.jpg720×540 104 KB


Slide2.jpg720×540 95.3 KB


Slide1.jpg720×540 26.2 KB

#4

Here are the other two figures. It seems that i can post as many as 5 images once.

Slide7.jpg720×540 55.9 KB


Slide6.jpg720×540 79.3 KB

#5

I’m sorry, I should have more clear: Can you upload some examples of the actual raw images that you are detecting objects in? If you like, you can post the pipeline as well.
-Mark

#6

Hi,

Attached are a part of the raw data under processing and the pipeline I used. the bright spots represent the nuclei with YFP signal which I try to segment, quantify and track in time course. Finally i aim to obtain quantitative traits of the YFP signal of individual cells.
p6YFP.cp (7.12 KB)







#7

Hi,

I’m attaching a pipeline that should deal with some of the problems you mention. A few comments:

  • I’ve changed the settings in the 1st IdentifyPrimaryObjects module to detect objects better. From the figures you posted, I can’t tell what makes the objects to be deleted different from the objects to be retained, at least on the basis of intensity alone. So I’ve done my best to identify them all.
  • Given the above point, it seems that Edit and IdentifyObjectsManually modules are still needed, so I’ve added module which should combine them for you.

Second, the error you are getting is a bug; the workaround is to not save outlines when you use IdentifyObjectsManually. We will fix for this for the next release.

Regards,
-Mark
2011_05_01.cp (11.2 KB)

#8

Thank you so much! I will try it.