Fibre analysis in microscopy images

[darenggli]

Hi all,

I am fairly new to image analysis, so I am unsure if what I am trying to do might be too much of a challenge for ImageJ. The only experience I have with the program is inserting scale bars to my images, but after watching some tutorials it seems like I should be able to get a lot more out of it.
My samples are fibres of various shape and sizes, but because of the sample properties they are really hard to image (image quality varies quite significantly from sample to sample). My goal is to separate the fibres from the background to measure size (mainly length and diameter). Basically, I just want to get a rough idea about the size distributions in my samples. I have thousands of images, so ultimately I will try to automate the process but currently my issue is to find a protocol to get relatively consistent results. Any suggestions on how to approach this would be welcome. Unfortunately, I have saved all of my images as jpg, so hopefully this will not be an issue.

Thanks for help.

[allopatry (Alex Hall)]

Though they aren’t cells, I recommend trying stardist or cellpose. The shape of the fibers and imaging modality are roughly similar to the expected inputs for these tools, so you may find some success there.

In traditional image processing, you’d want to probably first compensate for the image quality with a flat field correction, shadow removal, and normalize grayscale through the set of data. You may also consider segmenting the small white specks and using infilling to digitally erase them.

At that point you could work on an edge finder.to trace the edges of the fibers. Since the fiber thickness varies so widely, it may be difficult to use morphological operators to enclose the largest fibers… But if you are successful with just the small ones, you could at least label them and use DiameterJ (DiameterJ - ImageJ) or another measurement tool.