Inconsistent cell counting results on different machines

Hi everyone!

Could anyone help me?

We’re doing cell-counting and we’re working with 3 different computers using ImageJ.
On the same sample, the 3 PCs gave 3 different results. Mine gave 5 positive cells the other gave 71 positive cells and the third one gave 49 positive cells. The used threshold was the same with the same particular sizes.

So: same sample, same settings and different computers gave different results. Why? What could be the problem?

Hi @Retike, welcome to the forum!

Did you happen to try @ctrueden’s suggestion above to check the Troubleshooting page? This gives some reasons why you might get different results on different machines.

If that doesn’t solve the problem, could you provide some more details to help people understand what you’re doing, and what the results were? For example:

  • A sample image or a screenshot of the original image and the detected cells
  • The macro you used, or the specific steps you took to get the results

Hope this helps!

1 Like

Hi @Retike,
could you please upload some images and your pipeline/script/macro used to analyze them.

Thank you,
Emanuele

ok, sorry @tswayne tswayne I didn’t see your post
we wrote the same…:wink:

1 Like

Hello and thank you 2!

This is ours:

open/sample.jpg

process/s.background/light/50.0/ok

dirts: marking out/ctrl f

image/type/8 bit

file/save as samplebw.jpg/save

This is for finding the
positives:

image/adjust/threshold/130/apply

plugins/part.analysis/nucleus counter/min.:50,
max.:5000/ok

closing the image

This is for finding the all:

open again/threshold setting: 220/apply

  1. plugins/part.analysis/nucleus
    counter/min.:200, max.:5000/ok

And I show the pictures as well. We took them with AxioVision and saved them as jpeg.

These are from the same slide but different fields of view. We found the problems on the same field of view. This means that it should give the same result (of course). I just uploaded all because you asked for some images. Now you can see our works.:slight_smile:

Maybe we’re doing something wrong or just forget about something…No idea.:frowning:

Please record a macro and post it here. You instructions to try to reproduce the workflow are impossible to follow.

Thank you for interesting. Unfortunatelly I don’t know what “macro” would be. Printscreen pictures to show the menupoints in ImageJ I mentioned above?
Sorry.:frowning:

Is there any official help of ImageJ? Or this forum should be that?

Yes, there is documentation. You can start by looking at the Help entry in the menu bar. It is advisable to read that before attempting to do any analysis.
A macro can be recorded so you can reproduce everything you are doing manually (which I think could be the issue here: you might be doing different things in the two machines).

See the Macro Recorder guide for details.

4 Likes

Adding to @gabriel’s and @ctrueden’s advice, a good place to learn more about ImageJ is here.

After you record your macro, you can copy and paste it here.

Please also note which specific image you analyzed, and also include the results table or other output.

Then do the same for each machine that gives different results.

This will help a lot to diagnose the issue.

2 Likes