Segment different channels without overlap regions

Hi everyone,

I want to calculate the area for each structure in the figure linked below (green = mitochondria, magenta = muscle fibrils).

IFM_mitochondria.tif (3.1 MB)

When I threshold, there is always be some portion of each channel that occupies space of the other channel, although that would be “impossible” for a very thin section (the structures don’t occupy the same space - voxel size is big enough that this does happen on border areas, but I would want to exclude that). An aggressive threshold for each channel makes me lose portions of the image. A less aggressive threshold makes it that the sum of both areas is bigger than if I just draw around both.

How can I make a selection that takes into account the threshold of both channels and does not allow overlapping regions?

As a bonus, this image comes from a stack where I would want volume measurements. Is it possible to measure areas with the “orthogonal views”, so I would have also a cross-sectional area?

Another way to look at it could be knowing the long fibrils are cylinders (the structures in magenta) of a constant diameter (we could take the largest value of thickness seen along the z-stack) we could use channel 3 (magenta) to generate a mask/selection representing the cylinders/tubes and exclude that selection from channel 1 (mitochondria, in green). But whether this is easy/possible with ImageJ, might just be my wishful thinking :slight_smile:

The original file is here:

Thanks a lot in advance,