 |
 |
Hi, The tensor fitting is done with a least squares method (weighted, if you use the -w,--wls option). I'm not sure what the best reference to use for this tensor fitting is though - so hopefully someone else can help with this. As for the data, you need to supply a mask image when running dtifit. All voxels within the mask (those with values of 1) is analysed and those outside of the mask are not. It does not look in the diffusion data to exclude zeros or negative values. Also, FSL cannot deal with NaN values are these *must* be removed before any FSL processing is done. The one FSL tool capable of this is fslmaths, which can be run with the -nan option. All the best, Mark On 7 Aug 2013, at 05:52, Guo Ling (SBIC) <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hi, May I know how the tensor fitting is done in ditfit of FSL? Could I have the reference for it please? Also, if there are 0, negative or NaN values in the DTI data, will those values be ignored in the tensor fitting? Thank you, Guo Ling