If the algorithm converges appropriately there is no penalty in
increasing the maximum number of fibres to estimate.
However, the bigger this number, the more difficult it is to get
convergence - you would have to do various comparisons of results
with (very) different burnin periods to convince yourself.
If you have 15 gradient directions only however, you should certainly
stick to 1 orientation. 2 fibre populations require 9 parameters to
estimate - with 15 datapoints only, you will not find sufficient
evidence for the extra parameters.
Cheers
T
On 2 Jun 2008, at 12:42, Juergen Haenggi wrote:
> Dear FSL experts
>
> In bedpostx we can select the number of fibres modelled per voxel.
> Because
> our DTI data were acquired with only 15 different diffusion
> directions. This
> is in contrast to the 60 directions used in the Behrens et al.,
> NeuroImage,
> 34, 144-155, 2007 paper, where two fibres per voxel were modelled.
>
> Should we modelled only one fibre per voxel in bedpostx because
> there are
> only 15 different diffusion directions? Does results could be
> biased when
> modelling tot many (e.g. three) fibres per voxel?
>
> Thanks in advance
> Best regards
> Juergen
>
> -----------------------------------------------------------
> Juergen Haenggi
> Ph.D. (Dr. des.)
> Division of Neuropsychology
> Institute of Psychology
> University of Zurich
> Binzmuehlestrasse 14, PO Box 25
> 8050 Zurich, Switzerland
> 0041 44 635 73 97 (phone office)
> 0041 76 445 86 84 (phone mobile)
> 0041 44 635 74 09 (fax office)
> BIN 4.D.04 (office room number)
> [log in to unmask] (email)
> http://www.psychologie.uzh.ch/neuropsy/ (website)
> http://www.juergenhaenggi.ch (private website)
> -----------------------------------------------------------
>
|