Detection of myelin

[JDG210]

Hi Volker,

I ran your g-ratio macro on a few of my images, and I have been surprised by the results of the myelin detection: the area that is displayed in my images seem quite far from the myelin I actually see around the axons. At first I thought my image might be the problem so I tried with one with fewer axons that are more appart from each other, but I still have that issue (axon selections in green and myelin selection in blue):

Would there be a parameter I could modify to correct that issue? Or should I only use the macro on images with round, very well defined axons?

Thanks a lot for your help.

Best regards

JD

[volker-baecker]

I could reproduce the problem with the example images from the project. I'll try to find the reason for the problem.

[volker-baecker]

Please first make sure that the grow_regions.jar from

https://github.com/MontpellierRessourcesImagerie/imagej_macros_and_scripts/wiki/MRI_g_ratio_Tools

is in the plugins folder of your FIJI installation. You need to restart FIJI after putting the file there. Without it the tool works but the calculation for the myelin takes a long time so that it seems that the green and blue are in the same place.

This still does not explain the results I see on your screenshots, but to get ahead with this I need at least one of those input images.

Best,
Volker

[JDG210]

Hi Volker,
I installed the plugin and it does go way faster than before, and up to generation 1000 instead of 200. But the resulting myelin selection looks similarly inaccurate.
On my last try I noticed that the selection may be partly accurate and partly shifted... see below for selection #32 around axon #14:

Please find attached two images I tried the macro on, let me know if you can obtain something more accurate than me!

Thanks for your help, much appreciated.
01_5K.zip
.
11_7k.zip

Best

[volker-baecker]

Here is what I get. It doesn't look great, but the result is understandable. The myelin is searched around the axons. You could augment the filter size and with some manual editing of the axons it might be possible to get a reasonable result.

So I don't understand the result you got. I agree that the blue rois seem to be shifted, but I have no idea how this can have happened and it doesn't happen when I run it on your images. I am a bit out of ideas here.... Maybe you can try a clean FIJI installation with ImageJ 1.53k and make sure not to touch anything while the tool is calculating...

After that the only thing we could do is make a remote debugging session via zoom or a similar tool.

Best,
Volker

[bigzero61]

The same issue occurs to me as well.

The myelin rois shifted and every time I ran the 'f' command I got different results on the same image. The number of axons also spontaneously changed.

[volker-baecker]

Hi @bigzero61,
yes, I can now reproduce the problem of the myelin segmentation not working as expected. I'll look into it.

[volker-baecker]

Hi @bigzero61,

it should be fixed in the latest version. Note that (as written in the instructions), you need to select all axons in the image, adding those missed by the automatic procedure manually, for the global results to make sense.

Best,
Volker