Vesicle characterisation

[CMatti]

Hi all,
I try to characterise vesicle staining of two channels, one with protein of interest (0) and with a specific marker (1). What I am doing is using a spot detection to get my vesicles recognised in each channel, and then use labelling arithmetic to define, which actually do overlap (Setting AND). However, I realised, that it then also marks those which only partially overlap, which I don’t want. Unfortunately, I can’t use “Congruent”, as the signal is almost never 100% the same in both channels even when both vesicles overlap to 100%.
Therefore, I would like to work with the centre of mass (using Segment features and a math code), and then see the distance, and if it is below a certain threshold, count the two vesicles as overlapping. I know there is already a nice example workflow, however, this one calculates from one ROI to all other labels.
I would like to use the Labeling Arithmetic to get the ones which overlap somehow, and then use those new generated labels, to take always the two who led to that overlap and only calculate the distance between their centre of mass.
Can you give me some idea on how to start this? I guess I need something to iterate through the overlapping labels one by one, and in this loop use Labeling arithmetic with AND to get the corresponding vesicle. Then I would have to define the centre of mass and then append the second vesicle informations next to the first one and then calculate the distance.
I created a workflow, up to the point where I am now, based on a workflow once made by stelfrich and a simplified example picture with some circles (two overlap just by parts, two overlap completely).

If you can give me some help would be great, also, if my approach has some issues as well.
Best, Chris

Count_Overlapping_Vesicles_8.knwf (172.1 KB) a.tif (470.9 KB)

[CMatti]

Hi all,

I realised, that the Segment Feature Node combined with the Compare Segments actually does the job of comparing all overlapping structures, and via the maximum relative pixel agreement gives me a tool to sort out “false positive” overlaps.
Unfortunately it also kills some vesicles which overlap 100% with tubular structures, as the structure is longer and therefore the relative overlap is smaller again.

But, is there an option to get the label information again out of the “overlap keys” column? Idea is that I then can get further information about those segments.

Best wishes, and thanks in advance,
Chris

[imagejan]

@CMatti

is there an option to get the label information again out of the “overlap keys” column?

Maybe not exactly the same issue, but this discussion might help:

Compute overlapping area of different labels in labeling image

Hi @stelfrich, sorry for the delayed follow-up. Sorry if I haven’t been very clear in my original post. Judging from your reply, you understood exactly what I was looking for In the end, this was exactly the missing piece for me, thanks for reminding me of that possibility to generate the “fragments” of overlapping labels. At about the time of your post here, I also (re-)discovered the Compare Segments node, and it turns out that combining your approach with Compare Segments …

[CMatti]

@imagejan

Maybe not exactly the same issue, but this discussion might help:

It did indeed! This was kind of the missing link. I used this now to calculate the centre of mass difference, as I was now able to align two different segment tables with some more nodes in between.