Hi Stefan
> Thanks Saad for the quick reply!
>
> I think its utterly important to highlight this for the non-
> mathematician or
> the likes community, simply because people without the technical
> background
> go through painstaking trial and error trying to get the bvecs right
> and
> actually are just interested in analyzing FA - there is actually a
> real
> sparsity of cookbook information (well if one looks at the formula
> of FA one
> could figure out that the Eigenvalues are
> directionally cancelled out).
I think getting the bvecs right is worth the hassle!
> By the way, is there a way of (easily, a bit of a heresy - say an
> Excel
> sheet...) generating standard vectors for any given number of gradient
> directions or perhaps a link pointing to someone having gone through
> this
> before?
I don't think it's feasible in excel (although I'm often surprised by
that software!).
The problem of distributing points on a sphere has been tackled by
many people from many angles.
Various algorithms exist, and they generally differ in the cost
function (e.g. electrostatic repulsion between the points)
and the methods used to optimise the cost function.
I wouldn't recommend that you spend time writing your own code for
this - I think most MRI constructors went through this and generated
gradient tables that go quite high in the number of directions. These
gradient tables come in the form of text files that you could request
from your favourite radiographer and interrogate from excel whenever
needed?
Cheers,
Saad.
>
> Best regards.
>
> Stefan
>
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
+44 (0) 1865 222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|