Having difficulties with trabecular number calculation!

I am currently using boneJ to calculate the morphometrics from cores from the femoral head. I have calculated BV/TV, BS/BV, Tb.Th, Tb.Sp with no issues.

However, when I run connectivity (after purifying my image) I am getting abnormally high values for Tb.N - 20-30! I have looked in the BoneJ manual and in other forums, but I haven’t found the reasoning for this.

Am I missing a step? I do the following:

  1. Import stack of CT images
  2. Set scale
  3. Crop image around core (cylinder selection)
  4. Binarise image
  5. Purify stack
  6. Run connectivity
  7. Have very high values for ConnD!

slice114.tif (37.1 KB)

Tb.N is a distinct measurement from Conn.D that uses some geometric assumptions and Tb.Th and Tb.Sp, and sometimes BV/TV, rather than directly measuring topological loops as Conn.D does. Best not to get them confused: they are not the same. See e.g. this thread:

Between 3 and 6 you should smooth the trabecular surfaces. A 3D Gaussian applied to the greyscale image prior to binarisation, or 3D dilation and erosion after the binarisation, can help to get rid of loops of noise that get counted as topological handles and included in the trabecular count.

Conn.D is measured in units of mm⁻³ (trabeculae per cubic mm). Since your core is roughly cylindrical, you should use the Connectivity result divided by an estimate of your core’s volume (πr ²h) to calculate Conn.D.

This doesn’t seem that high for Conn.D. Remember it’s a per cubic mm measurement of things that are typically on the order of 10s - 100s µm. See for a broad view of mammals and birds https://doi.org/10.1098/rspb.2011.0069. Conn.D of 20-30 mm⁻³ is in the middle of the range and typical for smaller mammals, but maybe a bit high for larger mammals.

1 Like

Hi Michael,

Best not to get them confused: they are not the same

Ah, that makes much more sense. I think there’s been some confusion between myself and colleagues on this!

Thank you for the link to the related thread, it was really helpful.

Between 3 and 6 you should smooth the trabecular surfaces

Since your core is roughly cylindrical, you should use the Connectivity result divided by an estimate of your core’s volume (π r ² h ) to calculate Conn.D.

I will try both of these and see how I get on!

Conn.D of 20-30 mm⁻³ is in the middle of the range and typical for smaller mammals, but maybe a bit high for larger mammals.

This is for human tissue, hence my concern - but that is a great article

Thank you for such a comprehensive reply, it is really appreciated!

1 Like

For a broad comparative view of primate trabecular bone, see: https://doi.org/10.1098/rspb.2013.0172. Note that Ryan and Shaw use both Conn.D and Tb.N in their paper.