I have a couple follow-up questions to an analyses. I performed a fixel based analyses to investigate fiber density and cross-section in a clinical population compared to healthy controls. I have also attempted to use FSL crossing fibers approach using the ball and sticks methods which yields F1 and F2 (major and minor fiber population PVF metrics). Could someone please elaborate on the differences between the PVF metric that FSL outputs and the fiber density metric that mrtrix generates? Hypothetically, should I see similar results using the two analyses or are they generating different information?
A second, tangential, question: is there a way to overlay a standard white matter atlas to define specific tracts or should this be done in other software?
Thanks in advance for the advice,
They should be broadly similar - however there are differences in the models, the particular way the information is represented, and the approaches used to derive these measures, so they will differ (possibly quite substantially).
I’m not sure I understand your question… Can you tell us exactly what you’re trying to do?
When you analysed PVF from ball and sticks, did you somehow use FBA? Or use multi-fibre TBSS? If the later then I’d expect the result to be very different (see the discussion of our 2015 and 2017 papers for a discussion on the TBSS FA-based projection step and crossing fibre regions). Aside from the projection step, there would also be differences due to how ‘clustering’ (TFCE vs CFE) is performed.
Also, I’d expect the quality of your data (b-value and number of directions) to influence how different the results are.
Could someone please elaborate on the differences between the PVF metric that FSL outputs and the fiber density metric that mrtrix generates?
In terms of the metrics themselves, the primary difference is that PVF is a normalised measure: the non-linear fit constrains the sum of all partial volume fractions to unity. AFD is a proportional, dimensionless measure. T2 effects therefore influences them in slightly different ways.
Is there a way to overlay a standard white matter atlas to define specific tracts or should this be done in other software?
MRtrix3 itself doesn’t come with an altas / parcellation. If you can perform the registration step you can use a combination of the Overlay and Vector Plot tools in
mrview to get a visual overlay. But if you’re looking for a data-driven labelling of FBA results, this isn’t currently possible, since no “standard” white matter atlas takes crossing-fibre information into account.