Directionality of neuron fibers

Hi all (and @tinevez),

I’m trying to analyze the orientation of neuron fibers (see image below), and I think the Directionality plugin for Fiji should be able to do it, but I’m having a couple of difficulties. In the image below, it looks to me like the fibers tend to travel left-right more than up-down, and I’d like to quantify this and have an objective way of measuring this.

In Fourier mode, the analysis (judging by the Orientation map) seems to ignore a lot of the image – many fibers show no color, and the ones that do are clustered near the middle.

This is much less a problem in Local gradient mode, but it does still seem to ignore 1) the dimmer fibers, and 2) regions in the middle of thick fibers. The image below shows a cropped region of the orientation map output, where there is no color on the up-down smaller fibers, and only the edges of the thick fiber are colored.

Orientation map for EXP029-1

With regards to the dimmer fibers, can you please tell me how the analysis decides what is real signal and what is background? And do you know of a way that I could dictate what the low threshold is so that real signal is not excluded?

With regards to the thickness of fibers, you might notice that the left-right fibers are typically much thicker than up-down fibers, but the orientation map keeps much of the middle of the tracts un-colored. In reality the thickness represents many fibers bunched together. Do you know of a way to get the analysis to include the full fiber width? The issue here is not resolution - even with very high resolution images at 63x on a confocal, these fibers are too close together to be resolved.

Thanks very much for any help!
Ryan

Hello @rhamnett

Normally I would say that Directionality is not the adequate tool for what you are trying to do.
Directionality is meant to measure an estimate of the proportion of fiber-like structures that are aligned vs random alignment.
Think about a photo of uncooked spaghettis. You want to know what % of them have a common direction, vs random orientation.
So Directionality is not good at picking local orientation or any local features.

For this I would advise you to use the excellent OrientationJ package.
http://bigwww.epfl.ch/demo/orientation/

Tell us how it goes.
best
jy

2 Likes

Please excuse me for interupting however ‘Herbie’ has a very good technique to determine directionality in complex situations. I don’t have the ability to find a link for you but I am sure someone here can.
Bob

Thank you both for your replies. I’ve now had an opportunity to try OrientationJ and agree that it may well be better suited to my purposes. I wonder if you could help me with a couple of extra things please?:
1 - When using the ‘distribution’ measure of OrientationJ, do dimmer fibers contribute less than brighter fibers do? I.e. that the measure is weighted, taking intensity into account? Or do all contribute equally?
2 - In an image such as mine, can you explain what ‘energy’ and ‘coherency’ mean please? I’m not clear on what the energy and coherency maps are that it produces, and I’d like to appropriately set the minimum thresholds.
Thank you
Ryan

Hi
@rhamnett

1.Directionality
You have very nice discussions on the study of “Directionality” here:
Super interesting.

https://forum.image.sc/search?expanded=true&q=Directionality

https://forum.image.sc/search?expanded=true&q=OrientationJ.

Can you file a png image to measure the thickness of the fibers?
The method shown in the following thread should work:( Exclude fibers by size/thichness)

It seems to work for me, but your image is a jpeg. (The little picture)
Greetings