Comparison of CSD and DTI derived tracts

Hi mrtrix experts!

For developers, first of all, thank you for this amazing program! So simple to use, yet extremely powerful.

We have an interesting clinical case at our hospital. Stroke damaged part of the brain where arcuate fasciculus is, and I would like to compare arcuate fasciculus of the patient using dti and csd tractography. And then compare the results from the 2 tractographies.
Its not much, but for a student, it would be a nice little report.

After all the preprocessing, I have calculated tracts using tckgen with the mentioned algorithms, every other parameter was the same, and Ive put the -number as 1000. Then, i made tckmap of each tract. And did mrstats using previously calculated fa, and adc map from tensor2metric command.

My question is, for my purpose. Since stroke alters the diffusivity in the area, is this a valid comparison of the 2 different approaches for tractography? Or perhaps, do you recommend any other one?
I expected, and got better values with csd tractography approach of course.

Thanks in advance!

Thanks, it’s always nice to hear are efforts are appreciated :+1:

This is very much a research question, and as such is going to be dictated primarily by what actual question you’re trying to answer. Assuming you’ve done all the processing in a consistent and objective manner (ideally blinded to the pathology, but that’s not possible in your case given the obvious nature of the lesion), then technically any comparison is valid. Where things get messy is when trying to interpret the results - what does it mean when you see a difference, and can you draw meaningful and useful conclusions from those results that aren’t potentially confounded by all manner of other related artefacts?

Which brings me to my final point here:

What do you mean by ‘better’ here? If you’re comparing FA & ADC between patients and controls, I’d find at least the FA results very difficult to interpret reliably - the arcuate will be going through all kinds of crossing fibre regions, making the FA essentially meaningless (at least if you’re hoping to interpret it as some form of marker of white matter ‘integrity’ :confounded:). ADC might be more readily interpretable, but is also affected to some extent by crossing fibres. On the other hand, if ‘better’ in this context refers to clearer separation between patients and controls, then that’s a perfectly valid conclusion to report.

Just my 2 cents - others may very well have different opinions on the issue…