Porous connectivity + trying to understand the thickness analysis results

bonej
Tags: #<Tag:0x00007fd5432a0358>

#1

First of all, thank you so much for this pluggin, it is incredible how powerfull it is.

I have two unrelated questions, could you please help me ? I work on a stack of a porous material, realised by µCT . I’m interested in the porous proporties of this material.

My first question is, how can I get the porous connectivity from my stack ? I tried to trick the pluggin by applying an invert LUT process to my stack but, in any case, the pluggin give me the same connectivity whether the “bone” is in black and the “marrow” (void, pores, in my case) is in white and vice-versa.

My second question is, with the thickness analysis, if it “defines the thickness at a point as the diameter of the greatest sphere that fits within the structure and which contains the point”, why the thickness map of the “bone” overlap the thickness map of the “marrow” ? The void of my material seems more important that is trully is, doesn’t it ?

Actually, I try to draw the pore size distribution of my material but the resolution of my stack is around the mean size of the pores… So I press [h] to get the histogram of my thickness map (of Th.Sp) and thus the pore size distribution, it seems that I don’t get multiples of my voxxel resolution. And I don’t know why. My voxxel resolution is 0.33µm, but the histogram starts at 0.6µm. And an another large number of 0.722µm appears as well… I don’t get why…

Could you help me please ?

Thank you again,

Best Regards,

Sylvain


#2

You have to invert the pixel values, not the LUT. Edit > Invert.

This is a quirk of the Local Thickness implementation, which you can work around by checking the Mask thickness map option.

That’s not adequate to do a Local Thickness analysis. If your structures are less than about 5-11 pixels thick, your thickness result will be very unreliable.[quote=“Sylvain1, post:1, topic:3307”]
My voxxel resolution is 0.33µm, but the histogram starts at 0.6µm.
[/quote]

Thickness results are reported as 2× radius, and radius is measured in increments of a single pixel spacing. Your features are too small to measure at this pixel spacing.


#3

Also - I can’t see the images you posted, but the text description was enough.


#4

This was a bug caused by a Discourse upgrade. The problem was fixed yesterday, so the images should be visible again now.


#5

Images are now visible, thanks @ctrueden


#6

Dear Michael,

Thank you so much for your answers, it’s been really helpfull.

Unfortunattely, I understand that, in my case, as the voxel size (0.3µm) is close to the mean diameter size (0.6µm), I can’t use the thickness map.

But i’m not willing to let it go. Maybe I should try another software (like imorph ?), or will I face the same issue ?

A second option would be to decrease the voxel size ? To 0.05µm for example ? In that case, the structure would be more than 5-11 pixels thick, right ? Is it ok to do that ? Is there an easy way to do it ?

Thank you so much,

Best regards,

Sylvain


#7

Thank you ctrueden for your excellent work :slight_smile:


#8

You can try subsampling your data by Image>Scale… and setting the X, Y, and Z Scale values to a value >1. Note that you don’t get any new information by doing this and the precision of your measurements is still limited by the image resolution and original pixel spacing. You might get a better/smoother result by doing the subsampling on your original greyscale images rather than the binary ones, and by Gaussian blurring before the Scale operation.

If your object of interest is fractal in nature (e.g. cracks in soil) then at some point (of feature scale) you have to accept that things smaller than your sampling are too small to measure with this technique.

If you can live with measurements at a bulk, not local, level, you might like to try something like the MIL approach in BoneJ or some other techniques like autocorrelation that can work on low resolution greyscale images.


#9

Thank you very much for your help,

I wish you the best,

Best Regards

S.