Dear experts,
I am a neurologist working on ALS. My question regards the possibility to perform FBA on a single patient and then perform it again after a certain amount of time looking for signficant changes. The idea is to do it in a few subjects.
It is not clear to me if it is possible to obtain a Fiber density value in a specific fixel or coordinates. Is there a specific command for that?
Is it possible to do it on single subjects without pooling data?
Moreover, I would be interested in performing a fiber bundle analysis based on FA values as expressed here (the fiber bundle as a heat map with values derived from FA):
The FD per fixel is extracted at the fod2fixel step per subject.
So yes, but that would give you just the FD values for a single subject then indeed. If you’re after “significant changes”, you’ll realistically still need populations to compare (to).
See the post again in the mean time: there’s an answer in there that may help you!
My question regards the possibility to perform FBA on a single patient and then perform it again after a certain amount of time looking for significant changes.
Fundamentally the issue with single-subject testing is that we use permutation testing to generate our null distributions rather than parametric statistics, and it is impossible to permute a single subject. So my only advice here is to consider very carefully exactly what your hypothesis is, and consequently what data are necessary in order to capture your null hypothesis.
It is not clear to me if it is possible to obtain a Fiber density value in a specific fixel or coordinates. Is there a specific command for that?
The FD per fixel is extracted at the fod2fixel step per subject.
I think the question here may relate more to the user extraction of fixel quantitative values for specific fixels of interest. If that’s the case, then it’s important firstly to fully understand the fixel directory format. In this format, the FD values for all fixels are stored within a 1D image, and therefore extraction of values of specific fixels is equivalent to defining a fixel mask, which is itself a 1D image containing ones and zeros that specify which fixels you are interested in. Statistics for a specific set of fixels can then be extracted using mrstats with the -mask option.
Definition of that fixel mask is however a little trickier. We do not yet have the capability to manually manipulate fixel ROIs within mrview. You can however create 3D ROIs and use those to select all fixels within any selected voxels using the voxel2fixel command. You can additionally apply thresholds on FD / peak amplitude to select only fixels of a specific size. Achieving a level of specificity over and above that is however I think not possible with the current set of tools.
Hi, I’d like to perform FBA on 10-20 tumour patients.
The tumours would be located in the same region. How much would it affect the results if they are in slightly different regions?
or should that happen before? Right now, MNI space is not that important for the study, first I should be able to get results and if there’s time, then register to MNI. So would that be feasible?
Sorry for the naive questions…
Best regards and thanks for the help!