Difference between cellprofiler2.1 and 3.18

What is the main difference between the two versions of cell profiler2.1 and 3.1.8? I found that when “IdentifyPrimaryObjects” identifies the nucleus, the cell profiler 3.1.8 is not only less than the cellprofiler2.1, but also discards the part that is obviously the nucleus. I think the cell profiler 3.1.8 version is newer than 2.1 and should be distinguished more nucleus, not less? Or am I setting the parameters when using “IdentifyPrimaryObjects”?

Hi day0,

Though I might not be able to tell you the list of differences between both the versions since I am not familier with fine details. But the major differerence is 3.0 and later versions does 3D analysis (basically can give Z-stack images). Nevertheless in Identify primary objects there should not be any difference between both the versions (because I had used both). Please check you parameter settings once. It would be great if you could share the screenshot.

Lakshmi
Fujfilm Wako Automation (Consultant)
www.wakoautomation.com

Thanks a lot !



These are the parameters that identify the nucleus.
You must know which images are from cell profiler 3.1.8 or cell profiler 2.1.Is it possible that the default parameters are set differently in different versions or that the default parameters are set more strictly in cell profiler 3.1.8 and are optimized based on cell profiler2.1?

There’s definitely a difference in the 2-class Otsu method, not sure about 3-class – it might be worth reading through this github issue.

Hi day0,

Yes I got the images from both the versions. I definitely know that there would be a slight differences between the versions but the result should not change ideal (Always its advisable to use a same version for a project). The difference in your sample is very high. But my observation is this from your sample set. The way it calculates the smoothing scale is different as mentioned in the this github issue & the same is evident in your sample. Just carefully observe the smoothened images from both the versions it looks different (Smoothening is more in old version). Another point is the adaptive window size. These two difference could be a reason for your changes between versions. But, technically with respect to source code I am not sure about the changes. Just wondering in case you tried other thresholding methods?

Regards,
Lakshmi
Fujfilm Wako Automation (Consultant)
www.wakoautomation.com