Maximum Intensity Projection


#1

Hi,

I apologise for bothering. I am a beginner in using CellProfiler and my question seems to be quite “stupid”, because I have an issue with Maximum Intensity projection. I took images of my cells in three channels, ten Z-positions, and several positions in the well. These images are saved separately in tiff format. My aim is to make Z-stack projection of individual channels in individual well positions.

I have prepared a pipeline where I identified each channel and grouped data based on well positions in “Groups”. When I launched the pipeline in “Start Test Mode”, I saw only the first image of whole channel set after MakeProjection. Not Z-stack projection. When the pipeline contained SaveImages and I launch it in “Analayze Images”, I saw only the last image of the set in output folder in the end of the analysis.

I tried to make maximum intensity projection in ImageJ and it worked. Nevertheless, I have a huge dataset of images which I need analyse, and I want to use CellProfiler.

I also add several pictures of my shorten pipeline and images.

Could anybody refer me to any manual or show me where I did the mistake?

Thank you in advance.

Jirka


#2

I’m sure it’s not stupidity at all! It’s a bit hard to get started sometimes, but we’re glad to help. :smile: From what I can tell, your configuration looks ok since in Groups you have what looks like a reasonable number of z planes (10) for each image.

Only seeing the first image in each set in TestMode is the expected behavior- in test mode, CP goes one image at a time, so you’d need to run it 10 times to see your final projection.

When the pipeline contained SaveImages and I launch it in “Analayze Images”, I saw only the last image of the set in output folder in the end of the analysis.

I’m assuming you mean you saw only the 10th Z plane rather than the Max projection. Can you either upload your pipeline or show us a screen shot of your SaveImages configuration? I have a suspicion that you might be saving the ‘DNA’ image in SaveImages rather than the ‘MIP_nuclei’ image, which would explain that behavior. Please let me know if that’s the case. If you think that’s configured correctly, uploading a sample pipeline and a set of z planes (a single zip file is probably the easiest) would help us diagnose the issue.

Welcome to the forum!


#3

Hi Beth,

Thank you for quick response. I supposed that Im not able to see the results of projection in Test Mode; Ive probably read it in previous comments. For this reason, I focused on my pipeline again and maybe found my mistake. As you wrote about settings in SaveImages, I did not notice that the default setting in “Image bit depth” is 8-bit integer. Using this setting, images looked like without maximum image projection, blurred. Now, I change 8 for 16, and images look pretty good. Beginners’ fault of inattention. 

I would like to ask you about preparation pipeline containing MakeProjection. If I understand properly, I cannot make MakeProjection and identify cell components from these images in the same pipeline, can I? I want to identify nuclei and cytoplasm and quantify actin in these processed images.

My second question is about Point Spread Function correction. Since my images originate from epifluorescent microscope, I would like to correct them. Is there any possibility to apply PSF correction on images in CP?

Thank you very much for your advice.

Jirka


#4

If I understand properly, I cannot make MakeProjection and identify cell components from these images in the same pipeline, can I? I want to identify nuclei and cytoplasm and quantify actin in these processed images.

Correct. You’ll have to make one pipeline that just makes your projections, then another that takes the projected images and does IdentifyPrimary, IdentifySecondary, and some measurement modules.

My second question is about Point Spread Function correction. Since my images originate from epifluorescent microscope, I would like to correct them. Is there any possibility to apply PSF correction on images in CP?

It’s something we’ve discussed, but it doesn’t exist in the software as of right now. I’ve personally used Huygens in the past, and I know there are plugins that do it for ImageJ, but I don’t have enough experience with them to make a particular recommendation. If you are going to do it, my recommendation is to do it BEFORE Z projecting.