 |
 |
Hi Adrian dtifit, both through OLS and WLS, does not impose any non-negativivity constraint on the tensor eigenvalues. So these are not avoided. WLS takes into account the fact that taking the log of the signal (to fit a linear model) changes the variance of the measurements (smaller signals will have larger variance) and weights the measurements according to the variance in the estimation. See Salvador et al, Human Brain Mappin, 2005 for more details. Cheers Stam On 23 Oct 2014, at 17:49, Adrian Castro Toledano <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hello all, my name is Adrian Castro from Biomedical Technology Center in Madrid. When launching dtifit, and evaluating with and without WLS option, I obtained that the resulting FA images with WLS present more values higher than 1 that the without WLS. I supposed that by using WLS would avoid the apparition of negative eigenvalues, thus causing values for FA higher than 1, however I obtain the against. Please, could you clarify me if these results obtained are possible?? Or maybe am I wrong??? Thanks a lot for your help. Best Regards. Adrian Castro Toledano