How to count surface of these fibers

Dear members, I do have pictures of membrane bundles. I would like to calculate the surface of the membranes. It is to say, all fibers are cut (see attachment) and I want to calculate the surface of each fiber.
The picture is from a micro-CT in Tiff format. I first improve brightness contrast then I make it an 8 bit image. Then I set a threshold but then I start to get troubles. I can not mask only the internal lumen of the fiber so counting the number of fibers and calculating their surface is not possible. For those interested the number of fibers should be close to 2000.

Any help from this experienced forum would be more than appreciated. As I am just starting with Fiji ImageJ, please be patient with me.

topTest_0934.tif (366.1 KB)

Thanks in advance

@Fiji69fds

Just a few things…

  1. How do you want to define “surface of the membranes”? I’m assuming you mean area in this case… but then - Do you mean the area of only the donut shape? Not the entire circle from the periphery (including fiber center hole)?
  2. Do you want each fiber’s individual area calculated? Or can you pool them all together and then average based on count?

Just try to better define exactly what you wish to measure… this will help us better help you. Too - here are a few helpful links on Segmenting in Fiji that you might find helpful:

And also Scripting in Fiji (for when you want to batch process things):

Hello Fiji69fds,
I used just the Image > adjust > threshold - using the over/under setting. I started with 50-150 and adjusted for finer fit then Analyze > Analyze Particles for the following results.
Fiji69fds_Results.csv (42.2 KB)
Fiji69fds_Summary.csv (143 Bytes)
You can play around with the settings some to refine it.
Hope this helps
Bob

1 Like

Dear etarena and smith_robertj, thank you for your assistance in helping me out. I will take your advice and play around with it. If I encounter some more problems I do hope I may contact you again.

1 Like

I have tried for several hours but I seem not to succeed. Etarena what I want to achieve is to count all circles (open and closed (white)) and subsequently the cross sectional area of all the circles including the white/grey outline. I should add some additional information: the image is a tiff image and the scale is 12.5 pixels/mm. The diameter of the circles is approximately 540 µm and I expect a number close to 2000 fibers.
What I did so far is:
set image to 8 bit
set scale to 12.5 pix/mm
image=>adjust=>brightness/contrast to a minimum of 11 and a maximum of 66 followed by
image=>process=>sharpen However it is clear that many circles still touch.(I wanted to attach a file but apparently you cannot do that in a “reply”)
I used the technique suggested by Bob (image=>adjust=>threshold (default; Over/under; 58 -157; dark background)
then analyze => particles (restricted size between 0.15-0.60 mm² and circularity between 0.5-1) but only 200 circles were counted
When removing the restrictions of size and circularity, count increased to approximately 1200 circles far below 2000 and on top of that large areas were not counted. In the counted areas incorrect outlines of the circles were present.
I am sorry to bother you with these probably simple questions but I am a novice in image analysis and unfortunately have no access right now to a power user in my institution.
Anyway thanks in advance for your time and efforts
Filip

Maybe give this a try in CellProfiler? I’ve made a pipeline that does the trick - not perfect but there are a couple of DisplayDataOnImage modules that would let you filter based on various criteria if it’s including or excluding too many.

The basic steps are:
Invert
Inverted

Identify
Identify

Filter based on size and shape
Filtered

Overlay so you can see how it’s doing
Overlay

Here is the pipeline:

Fibers.cppipe (15.0 KB)

1 Like

Dear Anne, thank you so much for your help. Sorry for my ignorance, but what is a pipeline and how do I use it? Can I install CellProfiler on windows7 (32 bit)?
Best regards,
Filip

No worries, just head to www.CellProfiler.org and see the download instructions there.

Once CellProfiler is running, you can drag and drop the pipeline file into the left panel. And drag and drop your images into the panel on the right.

1 Like

Dear Anne,

Thank you for your clarifications and pointing me towards the download site. However, when I want to start cellprofiler I get following message

After that I CellProfiler starts. What did I do wrong and how can I cure this.

Thank for all your support,

Filip

You did not do anything wrong; that message refers to something from an old version of CellProfiler. As long as it starts, you should be fine!

1 Like

Thank you so much, that is a relief.