The resulting FA is not very realistic: it is almost binary with very high FA in outliers located around the skull, then typically WM voxels present an FA of 0.7-0.8. The rest of them are very very close to 0.0 (see figure 1). This is an effect derived from the model in #1 and the preprocessing in #2.
Can you think of a more appropriate procedure (maybe iterative) to get a more realistic FA?
The resulting FA is not very realistic: it is almost binary with very high FA in outliers located around the skull, then typically WM voxels present an FA of 0.7-0.8. The rest of them are very very close to 0.0 (see figure 1). This is an effect derived from the model in #1 and the preprocessing in #2.
Can you think of a more appropriate procedure (maybe iterative) to get a more realistic FA?