Hi Martin,
I'm afraid that this is a very difficult problem indeed, and still
unsolved in the research field. What makes it difficult is that
the motion introduces other artifacts which affect the ability
of registration to align things well. Also, aligning a single slice
independently to a reference 3D volume is typically very poorly
conditioned and the results are inaccurate. When put together
this makes slice-by-slice correction, like you need to do, quite
unreliable and inaccurate. So we haven't released any tools to
do this, although we have tried some methods in the past and
are currently working on an integrated system to deal with the
artifacts and the motion together. However, this is still a while
off in the future.
So at the moment I'm afraid that if you have enough motion
that it is obvious to the eye, then registration just won't work
well and your artifacts will probably spoil any data anyway.
Sorry that I don't have better news.
All the best,
Mark
Martin Kavec wrote:
>Hello everybody,
>
>I have quite a lot of DTI series of newborns, where a baby moved between
>acquisitions of successive slices, so the slices look like a "fan". There are
>no ghosting artifacts, though. In some cases it is so bad (poor baby was
>probably too excited) that I would have to skip most of the DTI volumes, or
>the baby moved during b=0 scan. I still have anatomical images, which look
>fine, so I could eventually use those as a reference.
>
>Do you have any idea, how can I correct for this kind of motion? Can FLIRT
>cope with is?
>
>Thanks a lot in advance.
>
>Martin
>
>
>
|