Deriving primary eigenvalues and transverse eigenvalues from diffusion data for tensor-based analysis

[Ivana]

Hi all!

I am comparing ischemic stroke patients at two time points after stroke: week 5 and week 25. In order to derive some diffusion parameters, I performed dwi2tensor and tensor2metric for all the patients based on the .mif files I had and generated FA and MD values. I will then compare two ROIs that comprise the lesioned hemisphere and the healthy hemisphere to assess changes and differences along the white matter fibers in both sides with regards to these parameters (my aims in short).

My question has to do with deriving more parameters from the dwi.mif files. As far as my reading goes, it does not make sense to report only FA and MD as most studies also report the primary eigenvalues and transverse eigenvalues to complement the other two metrics, or report the ratios of all these parameters.

I see I can add the -ad image part of the command to compute the axial diffusivity (AD) of the diffusion tensor (equivalent to the principal eigenvalue). Since the help in the Terminal does not seem to display more information about the -rd image part of the command, I want to double check that the rd image would compute the radial diffusivity? Would that be equivalent to the λ23 or an average between the medium and minimal eigenvalues? Also, how does radial diffusivity differ conceptually from the mean diffusivity MD - does MD provide information about the average magnitude of molecular motion in all directions, and rd provides information only about the medium and minimal directions?

Please correct me if my understanding is wrong.

Would you in theory recommend any other tensor-based metrics to be included in such an analysis?

Thank you very much and best regards,
Ivana

[jdtournier]

@Ivana

Since the help in the Terminal does not seem to display more information about the -rd image part of the command, I want to double check that the rd image would compute the radial diffusivity? Would that be equivalent to the λ23 or an average between the medium and minimal eigenvalues?

That’s odd, the help page for the current version (which should match what you see on the terminal), says:

• -rd image compute the radial diffusivity (RD) of the diffusion tensor. (equivalent to the mean of the two non-principal eigenvalues)

This matches the last of your two suggestions (average between the medium and minimal eigenvalues).

@Ivana

Also, how does radial diffusivity differ conceptually from the mean diffusivity MD - does MD provide information about the average magnitude of molecular motion in all directions, and rd provides information only about the medium and minimal directions?

The idea is that it reports on the diffusivity across the main fibre orientation. How to interpret that is however not necessarily very reliable. This article explains many of these issues.

@Ivana

Would you in theory recommend any other tensor-based metrics to be included in such an analysis?

Unfortunately, you’re probably asking the wrong people…
I’ve spent most of my academic career working on non-tensor methods – I’m not sure I’d be the best person to advise on this…

[Ivana]

Thanks a lot!

The command does seem to work now so porbably it was an issue on my end.

As for the tensor-derived metrics, ideally I will also include more sophisticated metrics in my research - the tensor analysis is more of a preliminary part for the purpose of a university project. The read you sent is indeed very good!

Kind regards,
Ivana