[Insight-users] A problem of Sampling in the Mattes Mutual Information Registration

[Luis Ibanez]

Hi Caojie,

The recommended number of sampling points is 3% of the total
number of pixels in the region that you want to register.

If you use the mask, then the number of samples should be close
to 3% of the number of pixels inside the mask.


Note that this is just a recommended value. Take it as a good
starting parameter, but be aware that you may have to tune its
value.

In practice, it is more critical to select a good value for the
number of histogram bins than for the number of samples.



Regards,


   Luis



----------------
曹杰 wrote:

> Hi,
>     This is my problem:
>     When I use the Mattes Mutual Information Metric to registrate,
> 
>     the fixedimage's size is 128*128*20----(PET),spacing is  (3.92,3.92,4.25) 
>     the movingimage's size is 512*512*20----(CT),spacing is  (0.98,0.98,4.25)
>     
>     How many sample points should I take to registrate?
> Since the fixedimage is PET images, and in each image the content of the image is a small part,most of the image is the background pixels whose intensities are 0 .   According to this situation,how many sample points should be the best
> and need I use the FixedImageMask? If using the mask,how many sample points should be taken?
>     Thank you!
>                                                       ITK user   Cao Jie
> 
> -------------------------------------------------------------------------------------
> 周杰伦在"第一现场"与你面对面 http://smspop.163.com/special/jay200408/index2.html
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users