Identifying Primary objects in Light images


I’m trying to analyze some light images captured at high magnification. These are CRL 2570 Jurkat cell images, which are round in shape.

I captured images on color camera and then converted them to grayscale using GIMP. The grayscale images look good in terms of Signal to Noise (Bright objects on dark background).

I’m attaching the pipeline I used along with couple of images. I’m not getting results on these. Please help!

Jurkatimages.cp (4.61 KB)

Hi Chai,

It seems that whatever conversion you are doing, it’s producing a near-binary image. You might almost be better off using a manual threshold of something like 0.4 in IdentifyPrimaryObjects, with “None” for declumping. I’m attaching a pipeline which does a better job with the images you provided by thresholding the overall image, masking, then thresholding the result again.

However, I don’t think this approach is robust. The initial GIMP conversion is creating disconnected cell fragments which get identified individually, which would be inaccurate if you are looking for a cell count. Perhaps uploading an original color image to see if CellProfiler can do something with might be helpful?

2011_07_15.cp (6.02 KB)


What modification of this pipeline would you recommend to identify objects from a brightfield image of similar suspension cells? I have images of cells with their brightfield and fluorescence components. many thanks.