Separation of particles

Hello, I am a total ImageJ beginner and do not know much about it so far. I try to separate lipid droplets in bluish microscopy images from each other and the watershed-tools do not work in this case. I recognized that it would help to sort out very small particles (small cells I don’t want to count) and the small lipid droplets differ from the particles I want to exclude because they are by far more round than the small blood cells and fibroblasts. Can I somehow exclude only those small particles (below a certain area threshold) which have a low circularity (below certain threshold)? I would be so happy if somebody could explain this to me!
Thank you!!

@anna2

To get you some better help… we still need some information from you:

  1. Can you share an original image file with us here directly or via a link to a file sharing site? That way - we can see your datasets… to get a better idea of what workflows/tools might work best.

  2. Please let us know the exact steps/commands you ran so far … that way we can see what you tried exactly and where to go from there.

It’s true that watershed works best with circular objects - but seeing your images will help us figure this out… MorphoLibJ might also be a good tool for you try.

1 Like

Tank you for your quick answer and sorry for my delay!

I tried to play around with brightness/contrast (ctrl shift C) and threshold (ctrl shift T) and tried this before and after converting the image into 8 bit. Then I used the Analyze --> Analyze particles with area exclusion below 1029 square pixels. This didn’t work so I also used Process–> Binary–> Watershed, which caused a separation of the lipidroplets into very small fragments. Then I installed the Watershed segmentation plugin, but it also failed to separate the droplets, but generated different objects instead.

  • This should mean “Thank you!”

Ok so the MorphoLibJ @etadobson recommended delivered these outlines, which are by far better separated than anything I got before. How can I calculate areas from these?

@anna2

How exactly did you generate this image? Perhaps @iarganda might have some good input on this…?

You could try using this image as input for the Morphological Segmentation tool in MorphoLibJ. You should be able to separate the objects in this case…

1 Like

I got this image with the morphological segmentation in MorphoLibJ with object image, 32 bit, tolerance 30000, connectivity 4.

@anna2 - then you can generate results directly there… as long as you are happy with the overlays, etc. So you can generate watershed lines - export that result… Read through that linked page that I gave you there for that tool - especially the tutorial videos.

Once you have your results image - then you can use Analyze Particles to measure the areas of each object.

1 Like

Sorry to jump late to the conversation. To add up on what @etadobson already said, you can also use the output from Morphological Segmentation in the form of label images directly with the plugins under Plugins > MorphoLibJ > Analyze.

1 Like

@etadobson and @iarganda thank you for your advice! Sorry for my delayed answer!
Is there a way to make a drawback which particle was assigned to which line in the results table? This would be great because have set up a macro which filters out lines from the results according to some thresholds…

Sorry for the late replay, @anna2. The pixels of each particle in the label image will contain the number you are looking for. So label #1 would appear in the results table under the “Label” column with a value of 1. Does it make sense?

1 Like

@iarganda Thank you for your patience with me! Sorry, what I was looking for is a way to see which parts of the pictures were selected as “one particle”, for example to see which of the “colourful bubbles” generated by the Morphological Segmentation Plugin belong to which line in the results table generated by the Analyze function.