Missing tracts and derivate measures

[osa.a...@gmail.com]

Hello everyone, 

I have a comment / question to do, more looking for help than offering I fear: I'm working with Diffusion imaging (HARDI) white matter material from stroke patients, who had a scan at several timepoints after stroke. Of course the damage that stroke causes makes sometimes impossible the reconstruction of tracts. I have some questions about the "how to go on" when this happens:

1) When we know that one tract is not destroyed (because we could reconstruct it in the previous timepoint, and it was not totally bad), but we cannot reconstruct it again because of the scan not being clean enough or other reason, what do you do? Do you usually try other softwares, or directly work around the variables of interest and assume they're now " = 0" (thinking of FA, for example). 

2) Some tracts cannot be reconstructed at all because a lesion breaks the tract completely but, is there a definition of number of fibers  / ROI traversed by fibers that you manage to define a "remaining tract"? If, let's say, we have the AF territory (which is rather large) cut by a lesion in the frontal area which prevents the whole tract to be reconstructed, but of course a good posterior section of the tract would theoretically be available, is there a rule of thumb to define when is it good enough to try other stuff, or would you declare it impossible?

3) When all you need is the variables, and you have past data of the same tract of the same participant, but there is one time missing (due to processing issues or bad scanning) would you do some restriction / calculation to "figure out" the data that is missing based on the past data, or would you define that as "probably destroyed bundle" and take the figures that would reflect that as good ones?

A lot of questions, but I hope someone can give feedback on these. Thank you very much!

Alberto