Yes, in fact, it does not work! However, for me, it is strange that I have big differences between single shell and multi shell fits. I have area in CC and other WM locations where FA is very very low... Maurizio >Hi >I am sorry I have no idea why that option sneaked into the released DTIFIT. It has not been implemented yet :) >Cheers >Saad > On 9 Nov 2018, at 15:26, Maurizio Bergamino <[log in to unmask]> wrote: > > Hi Szabolcs, > I have 98 directions + 10 B0. > The b-values are: 500, 1000, 2000, 3000 > The voxel was in the CC. But I have big differences in other part of the brain. > > And... another question: > I saw that in 'dtifit' there is the option: --kurtdir -> Output parallel/perpendicular kurtosis maps (for multi-shell data) > Is it better than -kurt option? > Maurizio > > > > >> Hi Maurizio, > >> What are your b-values? >> And also: where is this voxel? Which part of the brain? > >> Best, >> Szabolcs > >>> On Wed, Nov 7, 2018 at 3:36 PM Maurizio Bergamino <[log in to unmask]> wrote: >>> Hi Saad, >>> Thank you for the explanation. >>> Now, everything is clearer for me! > >>> Maurizio > > >>> Hi Maurizio > >>> The kurtosis model essentially interpolates your diffusivities (slope of log(signal) vs bvalue) to their value at b=0. This interpolation assumes that the log(signal) vs bvalue curve is a quadratic with a curvature that is controlled by an extra ‘kurtosis’ parameter. > >>> In DTIFIT the --kurt option assumes a single kurtosis along all directions (as opposed to, e.g. a kurtosis tensor). However, it turns out this model >underestimates the ‘interpolated’ parallel diffusivity, which explains the lower FA value. One simple alternative is to have a parallel and a >>perpendicular kurtosis (so two extra parameters compared to a standard tensor model) but this is not available in the released version of DTIFIT. > >>> I hope this clarifies the issue. > >>> Best wishes >>> Saad > > > >>>> On 3 Nov 2018, at 16:21, Maurizio Bergamino <[log in to unmask]> wrote: >>>> >>>> Thanks Manuel, >>>> It is the same thing that I understood. >>>> But I did not expect to have big differences when I fitted the same DTI data by using two different procedures, with --kurt and without --kurt. >>>> FA difference between, for example, 0.72 and 0.42 in the same voxel, is big. >>>> >>>> Maurizio >>>> >>>> >>>> >>>> Hi Maurizio, >>>> >>>> If I understand correctly, when you activate the --kurt option, you are fitting the kurtosis model, that is more suitable for multishell data. The >>problem is if you want to compare with other FA values from literature, then probably you'd like to split the data and just use dtifit to one shell >>without the --kurt flag. I hope this helps. >>>> >>>> Best regards, >>>> >>>> Manuel >>>> >> >>>>> Hello, >>>>> It is the first time that I use a DTI multishell acquisition. My question is related to dtifit. >>>>> When I use dtifit, do I have to use the option --kurt to fit my data? >>>>> Because, I used dtifit with and without that option, and I had big differences in FA maps. >>>>> For example, in the same voxel, I had FA=0.72 (without --kurt) and FA=0.42 (with --kurt). >> >>>>> Thanks, >>>>> Maurizio >> >> ######################################################################## >> >> To unsubscribe from the FSL list, click the following link: >> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 > > ######################################################################## > > To unsubscribe from the FSL list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1 ######################################################################## ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1