Dear Maria,
> I have a DWI dataset and have ran eddy correction on it. I was expecting the SSE coming from dtifit applied on the eddy-corrected dataset to have a smaller range than the SSE computed on the distorted dataset. I've used as input for dtifit that was ran on the corrected dataset the rotated bvecs that eddy provides. Has anyone experienced this? If so, should it have a smaller residual error once the dataset is presumably corrected and the tensor estimation is fit to it?
that would definitely be my expectation. You say that the SSE has a larger range. What happens to the average SSE? Was there lots of movement/eddy-currents in the dataset to start with?
Jesper
>
> Best
> --
> Maria
|