CP measures always the same first image

Hi,

with the latest release, images load fine and the pipeline runs. When exporting the data, all the measurements are the same and correspond to the first image loaded.

Could you upload a pipeline and a couple of images so we can see if we can replicate this behavior?
-Mark

Hi,

I up loaded the pipeline and some images. After running through the pipeline and exporting to spreadsheet, I get the same measurements everywhere. Also, when analysing, the windows do not change but stay on the first cycle. Hope this helps.
Marc
Images.zip (6.89 MB)
testPIPE.cp (4.09 KB)

Hi Marc,

I tried this and it works fine for me, displaying, measuring, and outputting differently for each cycle. The only changes I made were:
(1) In LoadImages, pointed the last setting (Input Image File Location), to Default Input folder. You had it pointing to “…/mergedWells”. Possibly the images there (on your machine) are identical?
(2) I added ExportToSpreadsheet with its defaults. I assume you removed it by accident before you uploaded.

David

Hi David,

actually I did not add export to spread sheet in the pipeline, as this never worked for me in the past. I run a pipeline and export afterwards. I’ll try adding it to the pipeline and see what it gets me. Thanks in any case.
Marc

Hmm, that’s odd. We would be happy if you report any errors or problems to us when you experience them.

Sorry if this is elementary, and just a speculation, but note that you should be using the “Analyze images” button to run your analysis, not the new Test Mode menu. Another user was confused by this, and might cause the “problem” you are experiencing.

David

Hi David,

I do press ‘analyze images’, I understand you have to ask the question. I have added export to spreadsheet in the pipeline and still get the same result. When I export the file using CP 1.0, I also get always the same measurement. During the analysis, all the images are opened and analyzed as should be. It seems that the software does not write the results to each line, but copies the first readout into each line.
Sorry for not reporting the problems with the export module earlier, it was no big deal. In CP 2.0, the spreadsheet was created just fine, but as I said, contains the same measurement. Funnily, the processing time does vary from image to image. I attach the csv, so you can have a look for yourself. I repeat that behavior with several pipelines.
Thank you,
Marc
100417testOUT__1_Image.csv (4.02 KB)

Hi David,

I have now run the CP 2 on a Windows partition on my Mac and it works like a charm. The problem seems therefore to be linked to the installation on Mac platform. Hope this helps.
Marc

Hi Marc,

Glad it’s working now, but I’m still puzzled about your pipeline exporting duplicate entries. Since it happened in CP1 also, I have to suspect something in your settings was incorrect. The CSV you sent shows that there are always zero Nuclei objects found - is that expected? The output you sent is from a different pipeline than I have, so if you are still interested in this, please send along the pipeline and I’ll take a look.

Cheers,
David

Hi David,

I have tried running CP 2.0 with several pipelines that have worked successfully for many months on CP 1.0. The result is always the same, whether I measure nuclei or not. The problem is reproducible with validated pipelines and exclusive to Mac platforms. Have you tried it on a Mac?
Marc

Hi Marc,
Yes, I have tested this on my MacBook Pro (OS 10.5) successfully. The attached results are from running the test.cp pipeline, and you can see that the DefaultOUT_Image.csv has different values for each cycle, and that the Rab5 file is reasonable too. Try running this pipeline and see what you get.

Also, you might try downloading the Mac binary from the website again, as we have made a couple of changes to the download lately. However, we haven’t changed the version number, so it won’t be apparently different, but one fix was a Snow Leopard bugfix.

David
DefaultOUT__1_Rab5.csv (16.1 KB)
DefaultOUT__1_Image.csv (2.22 KB)
testPIPE2.cp (5.18 KB)

Hi David,

I at last managed to have time to download the latest version of CP 2.0 and also run your pipeline: same as before. The output is always the same. I will try on another computer at work and will also ask some of our hakers to have a look, it seems very bizarre that your setup behaves well (the csv files you sent look good) and here it does not. I have python 3 installed, do you think this could interfere? The default version is 2.6 and I have to call python3 specifically when I run a script in the terminal.
Marc

Hmm, yes this is quite strange.

A few questions/suggestions for you, so we can diagnose this:
(1) Have you tried running the compiled version of CP2? We’re up to r9777, 2nd Beta now (cellprofiler.org/download.shtml) and have silently updated this, so you might try downloading and installing this again, if you have already.
Is there a specific reason you need to run from the source?
(2) What Mac OS are you running? 10.what? We have some Snow Leopard fixes in the works.
(3) We currently support python 2.5.4. So it could well be that python 2.6 is the culprit. Python 3’s scipy is not fully compatible yet, so it will not work. Try installing python 2.5.4
(4) So to be clear, it works fine on your Windows partition on your Mac? Using BootCamp or something else?

David

Hi,

I do use the compiled version of CP2 and I did an install just before running the pipeline you sent me, so I have the latest compiled version. I was just mentioning the Python3 in case it might affect something.
I’m running MacOS 10.5.8, I have not yet tried on my 10.6 machine. On that machine the default python is 2.5.4. I run Windows using Bootcamp and VMware, runs just fine. I am sorry not to be able to help you more. I’ll try next week to run pipelines on other Macs in the lab and see what’s coming up. Am I the only one with this problem?
Marc

Hi Marc,

Since you are using the compiled version, python has nothing to do with it. Please confirm that you are really running the compiled version (in the top window frame, it should end with “v.9777”). Can you try downloading an Example CP data set and run it (cellprofiler.org/examples.shtml)? And certainly try it on other Macs that you have access to.

Best,
David

Also, could you run CP on the failing pipeline, quit, then open Console.app and filter for “CellProfiler” in the upper right, and post the results? The behavior you’re reporting is very strange. I have no idea what could cause it. Does CP run the correct number of times (once for each image you have)?

Hello David,

I just tried the Drosophila example and it worked well, so it really seems something’s amiss with my pipeline. I’ll work on it and let you know if I get any wiser.
Marc