Better Identification of particles using CellProfiler + memory problems

Image Analysis
#1

Hello there,

I’m quiet new to Cellprofiler and currently working on a pipeline to identify and quantify particles on TEM-pictures. I faced major problems because the background is so noisy and my particles have low intensities. However I managed to develop a pipeline that identifies most of my particles and ignores background and “tinies” (small particles, that form as well during particle creation). I attached my pipeline as well as an example TEM picture. Still I have a problem when analyzing more than one picture. Since my background is so noisy and my particles are low in intensity i need to adjust my identification treshold for every single picture, what is with the amount I’m working with a little annoying. Is there a more elegant way to solve this problem? Another problem is, that my pipeline takes ages to work and actually stops working after the second or third picture out of memory problems and I couln’t figure out how to get it running. Last but not least I’m facing problems when saving my results. CellProfiler is overwriting my results of the first picture with the results of the second picture. I know that there is a way to manage this, but I didn’t find out yet. I’m thankful for any afford and apologize for the english, it has been a while since my last english lesson :slight_smile:

best regards
NW_Felina

I’m running Cellprofiler 3.1.8 on windows 8

particleIdentification.cpproj (652.2 KB)
AT_V211_DOE3_Probe4_20x_4b.tif (16.1 MB)

#2

Welcome! And nice English :blush:

I’m not at a computer so I can only help with the last question: CellProfiler is designed to handle huge experiments so it only saves the data you specify. When you say it is overwriting your data for each image, I suspect that actually you are referring to the results windows showing each cycle’s data? If so, that is perfectly normal because those are just for you to see what is going on during processing.

If you want to save particular output images, just use the Save Images module. If you want time save particular measurements, use the Export to Spreadsheet module.

Hope this helps!

#3

Hello and thanks for the reply!

My aim is to save for each image analyzed one output image (there I’m using saveImage) and an excel sheet for my measurements (Export to Spreadsheet). At the end of each run CP is saving the output image with the same name, so when reaching the second image CP overwrites the data of the first. I can’t tell if at the end of the complete analyzis CP generates the data for each image in the set because the program ends out of memory problems, but at the moment I just have the data of the last image analyzed of the set :confused:

#4

Ah, ok, then that is a problem with configuring SaveImages then! Hopefully someone else can take a look at your pipeline’s settings and see what’s going on.

#5

Anyone has one more idea on how to solve my problems with the pipeline? Any reply is welcome :slight_smile:

#6

This attempt isn’t perfect, but it does save images properly and works (somewhat) faster. You can build on this if you want to find the “tinies” or something else.

particleIdentification_02.cpproj (531.4 KB)

Are these the “particles” you are looking for? They look like viral particles. Mind if I ask what kind?

02%20PM
02%20PM1730×1728 1.53 MB

1 Like
#7

Thank you johnmc for your reply. Indeed, these are the particles I want to identify and yes, these are viral-like particles based on baculoviruses. I will try to get your pipeline running, at the moment Cellprofiler crushes each time I start it^^