Myelin fiber quantification

Hello,

I’ve been working with CellProfiler for about a year now on fluorescent images and it works great. Recently though I have taken on a project to count and measure axons and myelin sheaths in transmitted light micrographs of sciatic nerve (please see Osmium Stained Sciative Nerve Starting for original image). The myelinated fibers are darkly stained so with a bit of illumination correction I have been able to identify the axons as the primary objects with success.

My conundrum is that when I try to propagate out from the primary axon objects to identify the entire fiber I never capture the entire fiber . I’ve tried using an annulus method and it works ok but I lose a significant amount of information since the axon size and sheath diameter are then directly correlated by a fixed amount (Typical result seen in Myelin sheath by 40 pixel annulus image)

The thing is that the IdentifyPrimaryObject algorithm (Otsu Adaptive) clearly marks the myelin sheaths beautifully with yellow lines (please see…Primary axon identification result image). If I could only figure out how to propagate to the yellow line I would be all set since the problem is then just an ImageMath subtraction operation.

Apologies…in reading this over it is clearly getting a bit whiny.

Might there be a chance a member of the esteemed CellProfiler team could take a look at the three attached images and the working pipeline and suggest how I might identify the area traced by the yellow line as a secondary object?

Thanks so much for the outstanding package…and if you can solve this I’m good for a beer at Area 4 down the street from your lab.

Best regards,

John

John M. Lincecum, Ph.D.
Director, Discovery Biology
ALS Therapy Development Institute
300 Technology Square, 4th Floor
Cambridge, MA 02139

email: jlincecum@als.net
direct: 617-441-7253
Web: als.net







Axons and Myelin version 4.cp (8.03 KB)

I’m not a member of the team but my first inclination would be to

  1. identify your axons as you are now
  2. then, invert your source image using ImageMath, and
  3. use IdentifySecondaryObjects on the inverted image to identify your myelin sheaths

Have you tried something of that nature?