BoneJ Moment of Inertia output resolution issue

Hello Fiji users!

I am using Fiji to visualize scans of long bones (humerus and femurs) for my PhD research. I want to measure different anatomical parameters. To do so, I first align my bones with Moment of Inertia, then I binarize my image (Process>Binary>Make Binary), and finally I use Slice Geometry. It works pretty well, the only problem is that the alignment with Moment of Inertia makes me lose a lot of quality and it’s not great to measure my parameters afterwards (see attached pictures).

Does anyone have an idea to correct this?

Thank you in advance for your help.

Yours sincerely,

J. G.


Moments uses only nearest neighbour interpolation, so if there is a lot of rotation during the alignment step you can get ‘jaggies’.

That’s made worse if you have bigger spacing in z than x and y, which is typical in clinical CT. If the rotation is mainly around the long axis (because you lined up the bone’s long axis with the instrument’s rotation axis) then you don’t need to run Moments, but you may like to use Orientation to apply anatomical axes to your bone’s xy plane.

Thank you for your quick answer.
So you’re advising me to manually align the bone in z and use Orientation to set x and y?

Best regards,

J. G.

Manually (physically) align the (real) specimen in the scanner during image acquisition so that the bone’s long axis is parallel with the scanner’s rotation axis. That’s usually enough to get your proximo-distal anatomical axis sufficiently aligned with image z axis, and is what people tend to do anyway because it’s the convenient way to put long bones in the instrument. (except when you have an undissected bone still articulated and covered in muscles, and you are a bit more constrained by gross anatomy).

Then use Orientation to set the (mediolateral / craniocaudal) anatomical axes in the transverse (xy) plane. This rotates the principal axes with fine precision, without having to rotate the pixels. If you don’t really care about specific axes but just want some measure of eccentricity (e.g. Imax / Imin) then don’t bother with Orientation.

All right, thank you. I’ll be careful about my future acquisitions. For the scans at my disposal, I will use manual rotation because I still feel that it has less impact on the quality.


J. G.

Feelings are important, but you (as a scientist) must also measure the impact on your results and make a rational decision based on a written down, defined criterion. I’d be interested to know whether the jaggies that offend your aesthetic sense have any bearing at all on your numerical results.