Stack / movie and a segmentation based on a single image

Hi,

I have a general queestion. I was wondering if it would be possible with CellProfiler to apply a segmentation based on one channel onto a series of images from a stack or movie?
Imagine:
Channel 1 : *C01.tif is a single image which should be used for segmentation.
Channel 2 : *C02.tif are many image of a stack or movies to be used for measurement base on the segmentation of Channel 1 (the image from Channel 2 have tags such as Z01, Z02… Z10 or T0001, T0002…T0100 for 10 z-planes or 100 frames of a movie).

I havent found a way to do that in CP so far, and I dont know who to set up the NamesAndTypes and/or Groups module to solve that problem.

Solution 1 might be for few images, like 10 z-planes, to created the 10 image in the NamesAndTypes, but that works only for a low number of image in the series.

Solution 2 that would also work for larger image series, like 100 frames, could be to duplicate the Channel 1 image and make the file name consitstent with the series (take 1 image and make a 100 copies with file tages like T0001, T0002…T0100) using some other software (FIJI, KNIME, python …). Then each image of the series would have its counter part of channel 1 for segmentation. However that is a duplication of the data.

Both work, and I did it like that already few times, but it there a better or proper way of assigning one image to a series of images?

Of cause I have dozens / hundreds of these cases (wells and fields) and my the current analysis task I have 2400 images in the series, so I am a but hesitant to do the duplication of channel 1 as a solution.

Any help or ideas on that topic is much appreciated!

Many thanks!

Martin

This is definitely possible! You should check out the help for the “Image set matching method” in the NamesAndTypes module, which will explain to you exactly how you want to do this; I also recently explained this to someone looking to accomplish a similar task in the thread linked below. In his case, he was interested in making the objects separately, which you also can do (and might save time if both a) the identification of your objects in C1 takes a long time and b) you don’t need C1 for anything else, like measurement), or you can just have IdentifyPrimaryObjects as part of the same pipeline and it should return the same results each time.

We are also working on a tutorial video and blog post on this topic, as we know it can be a bit confusing to set up- stay tuned!

Thanks a lot Beth!
I always thought the metadata needs always a match (mostly I just use order in NamesAndTypes).
So the trick is that it is possible to just leave metadata unmatched (None), and then CP will assign one image to all images of a stack (here Z01 to Z03). I attached a screenshot of a small example.
Thanks a lot for the quick help!!
Martin

1 Like