How to extract the space between two different membranes

I am trying to extract and then be able to measure in 3D the space between two different membranes labelled with different colours. I have a z-stack acquired with a Lightsheet microscope, and I have managed to clean every channel so that when I threshold each channel I just obtain the membranes. I would like to know if there is a way I can extract the information of the layer between the membranes, since I am interested in measuring it.
Find attached a slice of a stack where you can appreciate how the two membranes look like.
TIFF_color3_Comp-236.tif (3.1 MB)

All the help is really appreciated!
Thanks in advance.

1 Like

It would be useful if you could describe what you want to measure about it and how precisely.

Perhaps what you are interested is the distance between the two membranes relative to a position on a manifold representing a membrane. Maybe you are interested in just the volume of the space to the precision of the sampled voxels. Perhaps you need additional precision and we should employ interpolation to resample the image.

Hi Mark!
Thanks for your reply!
Perhaps it is easier if I explain what the membranes are. The green membrane or layer is the myocardium of a developing zebrafish heart, and the magenta one is the endocardium. I am interested in knowing in which regions the layer between them is thicker. We can see it is thicker on one side particularly on the chamber you see on the right side, but would like to actually measure it in all directions and characterise it. The ideal thing would be to get the volume of the layer in between and then trace a centreline through the chambers (centreline of the blood flow) and measure it radially tracing perpendicular cross planes in different points of the centreline.
Don’t know if I made my self clear. Let me know if you have any additional question.

Hi Juliana!
I think what you have explained makes sense. In summary, you would have to:

  • Define your center
  • Define your radial lines, rotating them at same angles (ie. 0 90 180 270 degrees), for example.
  • Obtain the line profile of each radial line (either in each channel or if you have done a merge image. You would see two peaks, one belonging to the myocardium and one to the endocardium.
  • Measure the distance between maxima
  • Finally you would obtain something like Angle vs Separation.

If you have some experience, I think in Matlab or Python this would be very straightforward (but probably also very doable with FIJI scripting language).
That for a 2D analysis. Since you mention that your dataset is from Lightsheet microscopy, probably you would like to do it in 3D. For calculating the centerline of vascular structures, I have found super useful the Vascular Modelling Tool Kit platform (they even have a tutorial). Then, you would need to interpolated points along that centerline and repeat what I listed before.

I have done some similar stuff for some of my experiments, but I am currently in the process of writing/submitting my thesis. If you don´t find the solution before August, after that, I can try to adapt my script to your data, if you want.


1 Like

Hi Carlos!
Thanks for your reply.
Yes, exactly what you describe is what I want to do (in 2D and 3D). I have been looking at the web page of the VMTK platform and looks awesome! I have used Python before, so maybe this platform might be the best way to at least create the centreline through the heart and then analyse the thickness in the different radial directions in 3D.
While I work on that, I would anyway like some help trying to extract the volume of the space between the two layers, so that I can visualize it as a volume. I think that the best way this information can be interpreted is showing it as a thickness colormap and for that I really need the volume.

I’ll give the VMTK a go and let you know if I have any doubt.
Meanwhile, if you have any ideas for the volume extraction, that would be great.

Good luck with your thesis submission!
Best wishes,

Hi Juliana,

Some brainstorming here (probably there is much smarter solutions out there but maybe this helps a bit).

I was thinking that you could reconstruct/segment the lumen of both structures and then substract the inner one to the outer one.

For doing this, there are several ways (plus your tissue seems to be relatively simple). One could be to binarize your image and play around with morphological operations (erosion, dilation, close…) until you get a solid “cylinder”/“circle”. Then, substract one to the other (in FIJI, you have the “Image calculator…”). The resulting image would be a binary image of your space and you can calculate the volume by just summing all the voxels/pixel values.

Other alternative, could be using 3D Slicer (also open source and with a huge community behind) to segment both structures using seeding points. Furthermore, you could also merge your two channels and to the segmentation on this merged one (saving the substraction step).

You can follow this tutorial ( to learn a bit more on this task and the software (algo check their documentation and tutorials in their webpage).

I hope it makes sense and helps.

Good luck,