Looking to optimize analysis of a large live cell imaging dataset

Hello all,

I’m trying to set up a live cell imaging pipeline using the TrackObjects module in CellProfiler. I was hoping to request help with the following questions:

  1. Is there a way I can test an analysis pipeline (and the effect of changing specific parameters) with a smaller image set? i.e. visualize the tracks and somehow cross-reference with the imaging videos that I have to make sure that the pipeline is working well? If so, how would I do this?
  2. Does the community have any suggestions in terms of how I can get this tracking to work well?

I’ve briefly described the system below. Please let me know if I am missing any other critical information needed to provide insight.

I’ve attached representative images of one timepoint of the three channels that I acquired below. A larger temporal sequence can be found in this folder: https://drive.google.com/drive/folders/1SLjWGounzTbyAQTR4Y06c0VNh0swSrs4?usp=sharing . The folder also has the cpproj file that I have developed, a snapshot of it is copied below.

When I run this pipeline through the entire dataset, the results don’t make much sense and I’d like to test it with a smaller dataset and verify if the tracks being generated are reliable.

The images are of three channels. The first one stains the nuclei. The second and third channels can highlight cytoplasmic areas of the cells being imaged. The goal of the pipeline is to get intensity values associated with each cell over the course of the entire movie. As such, I’d like to be confident that the tracking software is working well.

The challenge as indicated before is that when I use this pipeline for the entire dataset, the results are a little dubious. But I don’t know how to verify that the pipeline is working well. i.e. be able to cross reference the tracks with the videos (for instance). Other strategies of cross referencing would be very welcome!

I am sincerely grateful for any suggestions/helpful guidance that the community may be able to provide.

Many thanks in advance!

r01c02f01p01-ch3sk01fk1fl1.tiff (3.0 MB) r01c02f01p01-ch1sk01fk1fl1.tiff (2.0 MB) r01c02f01p01-ch2sk01fk1fl1.tiff (2.8 MB)

Hi @mtTriesCellprofiler,

To address your quesions:

1 - You could quite readily only load a limited set of images into the interface. Just clear the file list and only add in a subset of files. Alternatively, if you only have a single file containing everything, you might want to use the NamesAndTypes filters to limit how many frames you’ll analyse. Another option would be to add a Crop module at the start to restrict analysis to a small part of the original image.

2 - It’s difficult to give precise settings for tracking itself. I’d recommend trying the ‘overlap’ and ‘distance’ tracking methods before the more complex LAP tracker. If using those methods you can use the SaveImages module to save a copy of the resulting segmentation, and combine those into a movie to test how well the tracking worked. With LAP this is not possible as tracking labels are assigned at the end of the run. However, CellProfiler Analyst Tracer was developed to allow you to visualize tracking data from LAP experiments.

Hope that helps.

1 Like

Hi @DStirling
Thank a ton for your advice! I’ll try that out and get back to you if things don’t work. I had posted another message just before this - please disregard that. I didn’t read your last message carefully enough. You have answered my question in your post already.
Many thanks!!