[Insight-users] Resampling and MutualInformation

[Lars Nygard]

Luis,

Thanks for your suggestions, I changed my interpolators and 
stopped to resample the fixed image and now my metric values I
get with the registration problems are close to the values I get when
registering them direct.
There's one thing though... I have five mri image (t1, t2, flair, dti and a
mr angiography). Ive created some problem by translating and roting 
the images(10 degree and 10 milimeter) and I register them all to the t1 image. 
All registration are good however when I rotate 10 degrees around x-axis and
translate -10mm along z-axis I get an additional y-translation (about 3mm every
time) and the z-tranlation is 9.4 (where it should be 10)
So I register t2, flair, dti and mra with rotation 10 degree around x-axis and 10mm 
along minus z-axis and in all cases I get the same error in y-translation.
The image matrix are 256x256x180, 1mm spacing. 
I first thought that it might be another representation of the same transformation but
my error indicates different (it's about 3mm, where it's 1mm in all other problems)
If you have any idea about what's causing this Id be very curious. I was thinking it
might have anything with the resampling to do. Other than that Im in the dark.
thanks for any ideas,
lars

----- Original Message ----
From: Luis Ibanez <luis.ibanez at kitware.com>
To: Lars Nygard <lnygard at yahoo.com>
Cc: insight-users at itk.org
Sent: Thursday, September 20, 2007 5:19:29 PM
Subject: Re: [Insight-users] Resampling and MutualInformation


Hi Lars,

Resampling will indeed change the values of Mutual Information,
in particular depending on:

   1) What interpolator you used during the resampling and
   2) Whether you resamples the fixed or moving image, and
   3) How many spatial samples you are using

Note that "higher" is ambiguos here. What exact class are you
using as metric ?  (MattesMutualInformation ? MutualInformation ?
another one ?).  Some of them report the negative of mutual
information.

How much was the change ?

A log of the output produced by the optimizer observer may
be illuminating here....


   Please give us more information,


      Thanks


        Luis


---------------------
Lars Nygard wrote:
> Hi,
> 
> Does anybody know how or if resampling an image affects
> the MutualInformation?? I have resampled my image to a 
> greater size so that image content is mapped outside my
> image during rotation but now my values for mutual information
> changed (they get higher, so less correspondence)
> thanks for any ideas,
> lars
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
> 





      
_________________________________________________________
Alt i én. Få Yahoo! Mail med adressekartotek, kalender og
notisblokk. http://no.mail.yahoo.com