Trabecular segmentation

Hi! I need to use Auto Threshold Optimizing, but I dont know if it is still available. in BONEJ. Thanks!

We removed Optimise Threshold from BoneJ because most of the time the autothreshold included in ImageJ is sufficient, and the title word ‘optimise’ may have been giving users a false sense of security.

If you really need a threshold guesser based on minimising the connectivity vs. threshold value function, which was the special option included in BoneJ’s Optimise Threshold, it may be possible to write it in a simple script or macro.

Thanks for writing me.
What I need is to change the cortical bone to black, thus only trabecular bone is analyzed.

How can I do it?

You need to segment cortical from trabecular bone. There are a couple of approaches in the literature. One way is to run Thickness, then threshold on thickness and fine-tune the binary mask to delete the cortex from the input image.

Im sorry, but I didnt understand what you explained to me.
I need to analize Tb.Th, Tb.Sp, BV/TV and connectivity of trabecular bone.

For this I need to remove cortical bone from the image.

I think that what I can do is a Threshold that alows me to visualize the intermediate tones (gray of the cancellous bone) and that both the white of the cortical bone and the black of the air, disappear. Is that correct? and is that possible?

Or there is an other way to analize Tb.Th, Tb.Sp, BV/TV and connectivity of trabecular bone?

Thank you very much

If this is possible it’s likely that you don’t have sufficient resolution to measure trabeculae. Would you mind posting an image here? What is the pixel spacing in xy and z?

It is a 3D tomography that uses 200 micron isotropic voxel.


I have to calculate TbTh TbSp BV/TV and connectivity. And I dont know hoy. I need to evaluate only trabecular bone.


Libre de virus.


You don’t have adequate resolution to measure trabeculae, sorry. Even the largest trabeculae (0.5 mm diameter) will be resolved by 2 pixels in the best case, with a lot of partial filling artefact on the edges. You might be able to to get an OK estimate of BV/TV, but you’ll need to take special care about your segmentation, or even use a greyscale approach, because so many of your trabecular pixels are only partially ‘inside’ trabeculae.

I want to
do a similar study to the one I sent to you. I copy two paragraphs with the
most importnat information. They use 250 µm.

thickness(Tb.Th), trabecular separation (Tb.S), bone volume fraction (BV/TV),

surface (BS/TV) and connectivity density were calculated.

CBCT imaging

CBCT was performed using WhiteFox_ unit (ActeonGroup, Milan, Itay). The field
of view (FOV) and the voxel size were set at 13 9 15 cm and 250 µm respectively.This FOV allowed scanning both jaws with
their basalbone. The tube current, tube voltage, and exposure timeare preset at
9 mA, 105 kv and 9 s respectively.

Don’t. You can’t measure things that are smaller than the precision of the instrument. This paper should not have passed peer review if it reports Tb.Th and Tb.Sp with a pixel spacing of 250 µm, measured with a sphere fitting algorithm.

What if I used 75µm?

Libre de virus.

Depends on the thickness of the trabeculae. You need at least 5 pixels across every trabecula. 75 µm might be OK for elephant, but marginal for most other species.

Ok. Thank you. Now I am doing a sheep bone analysis, but mainly I need to analyze human bone. Aside from image resolution (which should be improved), my first question was regarding how can I analyze only trabecular and no cortical bone in images in ImageJ.

I reed an study that did this:

Histomorphometric analysis

Trabecular bone histomorphometry was performed using ImageJ 1.47v (Rasband, W.S., ImageJ; US National Institutesof Health, Bethesda, MD, USA) with BoneJ (version 1.3.11) plug-ins. After opening slices in ImageJ as a virtual stack, BoneJ developers’ instructions19 related tobone histomorphometry were followed. Briefly, thresholdswere adjusted using “auto thresholds optimizing” (in theBoneJ dropdown menu). Tb.S, Tb.Th, volume fractionBV/TV, specific bone surface (BS/TV), and connectivity density were calculated (using BoneJ dropdown menu).BoneJ does not calculate Tb.N. Optimizing thresholds changed the cortical bone to black, thus only trabecular bone is analyzed. For Tb,Th and Tb.S, the mean, the standard deviation (SD), and the maximum values were recorded and used in the statistical analysis.

You told me that Auto threshold optimizing is not available any more. How can I do it?

Just use ImageJ’s Image>Adjust>Threshold, click stack histogram and one of the inbuilt autothreshold options.

Bear in mind that these autothreshold algorithms use the image histogram as input so they are sensitive to the relative amounts of foreground and background in the image.

Ok. Thank you very much.