[Insight-users] Is there a way to quantify accuracy of a deformable registration?

Ganesh Narayanasamy nganesh76 at hotmail.com
Wed Apr 28 12:58:31 EDT 2010




One of the assumptions made in image registration is that the true deformation is known to us. However, this is not the case in real world problems. How would you quantify the accuracy of a registration when the true deformation is not known.

In such a scenario, the widely used method for quantifying registration accuracy is by estimation of mean registration error (MRE) or true registration error. Pick a set of corresponding pairs of control points (or fiducial markers) on known structures or anatomy that can be seen on the original fixed and the moving images. The registration transformation applied to this set of fiducials in the fixed image volume gives the location of these reference fiducials in the moving image space. The mean Euclidean distance between these transformed fiducial points and their corresponding points selected earlier in the moving image volume gives the MRE.
This method is dependent on your or the readers ability, skill and patience to pick the right points and hence prone to human error.

Hope that helps,
Ganesh

************************************************************************

"What the mind of man can conceive and believe, It can achieve" - Napolean Hill

-Ganesh 
Narayanasamy,Ph.D., PostDoc-University of Kentucky Cancer Center

Earlier work: 
http://www.sciencedaily.com/releases/2007/05/070504133017.htm

************************************************************************

 		 	   		  
_________________________________________________________________
The New Busy is not the old busy. Search, chat and e-mail from your inbox.
http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_3
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20100428/87420588/attachment.htm>


More information about the Insight-users mailing list