Hi Colin,
The current implementation for multi-shell ODFs requires that the sequence 0, b1, b2, b3 is an arithmetic progression (i.e. that x_i-x_(i-1) is constant). None of the sequences you indicated fulfill that. For more info see Aganj et al, MRM 2010.
This a requirement that makes some computations analytic and the problem computationally tractable. That will change at some point in the future.
Notice that multi-shell ODFs are sharper than single-shell ones, therefore more prone to noise. Going to very high b values might not be a good idea, unless you have access to really strong gradient systems that give you some SNR at b>3000-4000.
Cheers
Stam
On 17 Aug 2012, at 22:51, Colin Reveley wrote:
> For multishell data Qboot requires bvals to have an arithmetic progression.
>
> Would 3700 6700 9700 work?
>
> How about 4000 7000 10000?
>
> Neither progression starts at 0.
>
> Obviously these are both lies, the first is closest to truth.
>
> If not I guess 3 6 9 but the true upper bval is 10500 about. I don't know enough to know if that would work out. It would be cool to try the data in fsl though. It actually has 5 shells but we knocked out the lower ones to try this. A few grads only.
>
> Thanks Colin
|