Note the areas with high diffusion anisotropy (white matter). Are
there any areas where the FA is greater than 1? Why might this happen?
answer
FA (Fractional Anisotropy) is equal to the variance of the 3 tensor eigenvalues, normalised to take values between 0 and 1.
When all the tensor eigenvalues are positive, this is indeed the case.
However, if one of the eigenvalues is negative, then FA can be higher than one.
In theory, a negative eigenvalue means that there is an orientation along which the diffusion coefficient is negative.
This is physically impossible. For example, it means that if we were to
measure diffusion weighted signal along that direction, then more
diffusion weighting increases the signal (instead of decreasing it). This can happen in practice, e.g. if poor SNR or head motion.
In DTIFIT, the tensor is calculated with no positivity constraints on
the eigenvalues, so situations where FA>1 may happen in practice.
No comments:
Post a Comment