[Insight-users] registration of images with missing data

[Matthias Schabel]

I'm interested in performing image registration on float/double  
images in which NaN entries indicate missing or bad data. I guess  
this falls into the more general sphere of registering images that  
have been masked - I wasn't able to find any information on this  
particular problem in the User's Guide. Any suggestions on how to do  
this efficiently? Specifically, I have two 3D images - one is a map  
of T1 measurements and the other is a map of mean intensity. In some  
cases it is not possible to get sensible values of T1 from the  
algorithm we are using, in which case the entry is set to NaN. The  
mean intensity map has no missing entries (but, in principle, I might  
want to register with an image having a different mask). Clearly,  
interpolation will potentially create 3x3 holes around each NaN  
entry, but other than that, I think the actual algorithms should work  
fine if there is a way to tell the image metric calculation to ignore  
NaNs...

Matthias

----------------------------------------------------------------
Matthias Schabel, Ph.D.
Assistant Professor, Department of Radiology
Utah Center for Advanced Imaging Research
Center for Advanced Medical Technologies
Room 2120
729 Arapeen Drive
Salt Lake City, UT  84108
801-587-9413 (work)
801-585-3592 (fax)
801-706-5760 (cell)
801-484-0811 (home)
matthias dot schabel at hsc dot utah dot edu
----------------------------------------------------------------









-------------- next part --------------
Skipped content of type multipart/mixed