[Insight-users] Picking a transform - costs

[N.E Mackenzie-Mackay]

I am currently trying to code 3d registration program that will 
register a 2d ultrasound image to a 3 dimensional ct set.

I am going to use metric: mattes mutual information - using a multi 
resolution approach
			    Interpolator: linear
			    Optimizer: Gradient Decent
			

The problem is I am confused on what transform I should use.  The 
example applications tend to use the QuaterionRigidTransform.  Coming 
from a graphics background I feel more comfortable using 
Euler3DTransform but am not sure of the difference in performance cost.

If anyone had some insight into my dilemma I would be grateful.  Also 
any suggestions about my project in general is welcome.

Thanks,
Neilson