Dear Ed,
You are correct that eddy_correct works volume by volume.
If your SNR is really low then it will probably be difficult to get
a very reliable registration. It may be helpful if you could
register images together with "similar" gradient directions (as
the differences due to diffusion changes will be less) and
then create averages of these that might have better SNR.
You'd need to script this yourself, but it might provide you
with a reasonable solution in very low SNR data.
All the best,
Mark
On 8 Mar 2012, at 13:07, Ed wrote:
> Dear Michel,
>
> Is is impossible to change the acquisition, so your second suggestion would be a solution. However, I don't understand this. The eddy-current correction tool operates on each volume separately by matching each volume to a reference volume, i.e., the B0 volume. So inserting average volumes will not result in a better match for the invidual volumes. Moreover, taking the average of volumes which are displaced from each other will make no real sense: the average will be blurred.
>
> Regards,
> Ed
>
|