Module to measure myelin sheath thickness

Dear all,

Does anyone of you have experience measuring the diameter and area of (transversally section) dendrites and their myelin sheath in electron
micrographs?
In our otolaryngology-laboratory we calculate these by circumferencing them manually in ImageJ, which is very time consuming (and my job while doing this research internship :wink: ).
Since the dendrites are easy to identify from the background because of their dark myelin sheath, I hope it would be possible to measure the size of axoplasm and thickness of myelin sheath in a more automatic way…

I tried to make a new module in ImageJ, but I’m experiencing a lot of difficulties.
First problem is the size of the images (±30Mb). They consist of 4-5 electron micrographs which overlap and for that reason are put together in photoshop. In CellProfiler it takes more than a day to upload them (on the Mac I use). I don’t want to make the pictures smaller because then I lose a lot of pixels (and information).
To try and make a module before resolving the size-problem, I tried to make a module with some decreased test-pictures, but I can’t even get the module to outline the primary objects. There seems to be to many background ‘fog’.

My idea would be to identify the outer (dark stained) myelin sheath as primary object, then identify the (light) axoplasm as secundary object, and then substract these from eachother. Can anyone help me with the start of this module?

In the end I’d also like to measure the packing density by dividing the total number of dendrites by the area outlined by the bony boundaries, but I might save this one for later…

I’d really appreciate your help, since I’m new to CellProfiler (and cell counting and research and module-writing in general…)

I’ve added an example of an (resampled) electron micrograph.

Thank you in advance.

Laurien Waaijer
Medical Student
Utrecht University, The Netherlands

Attempt #2, with files in JPEG…




Hi Laurien,

Thanks for giving CellProfiler a try. As it turns out, the mosaicked electron micrographs are far too large to be loaded into CellProfiler. However, I’ve worked out a tentative pipeline that might still be of assistance. I’ve tested it with a single micrograph, cropped to remove the caption label. Please note that some of the module settings still need to be optimized; see the annotations at the top of each module. Also note that if you use the single micrographs, there will probably some areas where the axoplasm/dentrites will hit the image edges and will not counted or segmented properly.

Hope this helps!
-Mark
2011_05_06.cp (10 KB)