[Insight-users] anisotropic datasets

[A.R.Lopessimoes at ewi.utwente.nl]

Hi,

 

I have two highly anisotropic MRI datasets (EPI T2* and T2) that I would
like to register.  After having read the section on the ITK Software
Guide about how to resample an anisotropic dataset to make it isotropic
(particularly the comments that accompanied the code lines), I started
wondering about which approach to follow.

 

My initial idea was to do this three-dimensionally, despite the poor
resolution on the z direction. Indeed, most papers dealing with the
registration of such datasets use normally 3D rigid registration.
However, the ITK Guide left me thinking whether there is any advantage
at all of doing it in 3D, since there will always be information missing
on the z-direction (even after the resampling to make the datasets
isotropic). On the other hand, slice-by-slice registration seems too
little, because after all we have several acquisitions on the third
dimension.

 

Since I am new to medical image registration, I was wondering if someone
could give some hints or further advice about how to handle this sort of
datasets. How "wrong" or inaccurate is it to perform 3D registration
with very "2D-like" datasets? And is it very different to make both
datasets isotropic and then register them or registering them without
making them isotropic?

 

Thank you in advance,

 

Rita

 

 

 

 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20100827/480c2465/attachment-0001.htm>