Confusion about BoneJ Thickness results

Dear imagej-user,

I am working on my soil CT images recently. I want to use (BoneJ>thickness) to get pore size distribution of soil sample. But I have a confusion about the results. My images are a stack with many slices. In order that I can better state my problems, I attach some pictures.

This is my original image, a 400 * 400 * 400 voxel image selected after contrast adjustment and filtering(Fig.1). The voxel is 5.2μm.

<Fig.1>

After binarization(Fig.2)

<Fig.2>
8

I went to bonej’s website to read the description of thickness, and it mentioned that “The plugin assumes that trabeculae are the foreground (255) of a binarised image”. So I think I need to invert the image to make the pores become foreground, like this(Fig.3)

<Fig.3>
9

Then, I run the Thickness, get the result as follows(Fig.4-6)

<Fig.4>
4

<Fig.5>

<Fig.6>

The result is confused with me.

The pore sizes are the pixel values of the image. This result means the max pixel of the pore in the images is 8 pixels.

But Thickness plugin, which defines the thickness at a point as the diameter of the greatest sphere that fits within the structure and which contains the point.

I wonder if there will be any mistake in the results? The pore size is really too small…I can feel the pores in Fig.4 I attached at least not so small. There must be more than 8 pixels…(I think the color should represent trabeculae, i.e. pore)

Also, I make the thickness with a non-invert image(Fig.2). The max pore size is 77 pixels. I think the result is more realistic…

But I think it’s logical to use the image of invert according BoneJ’s website.

I’m not sure which one is right… So I’d like to ask why the results turn out like this, and if I did something wrong?

Thanks in advance.

Best regards.
Mengying

Would you mind sharing this image for testing and feedback? Zipped tif stack is usually quite small.

Hi @mdoube
Thank you for your immediate reply. No problem, I can upload it.all-otsu-roi.rar (22.7 KB)
Because I am new in ImageJ, I use a small number of images to learn. The above result data comes from 15 slices.

By the way, I want to ask that use the invert images to do thickness is right?

Thank you very much for your help!!
Mengying

The thickness will be measured on the pixels with value = 255, regardless of the colour they appear on the screen. Hover your cursor over the pixels to see the pixel value in the ImageJ status bar.