[Insight-users] Re: One+One_Revolutionary

[Luis Ibanez]

Hi George,


These repeats on the Metric values are normal for the 1+1 optimizer.


They reflect the way in which this family of Evolutionary
optimizers work.


Evolutionary optimizers bred a population of points in the
parametric space. At every iteration, the population is
renew by mixing among the best individuals, and removing
some of the ones that are not well fitted.

The values that the optimizer reports at every iteration is the
one of the best fitted individual among the current population.
It is therefore common to see the same value appear in multiple
iterations until a better fitted individual appears in the
population.

The fact that you see the same value in several iterations is not
a problem. The population of samples in changing internally, and
simply has not found a better fit yet.

For detail on the rational behind this optimizer you may want
to look at the references indicated in the Doxygen documentation:

http://www.itk.org/Insight/Doxygen/html/classitk_1_1OnePlusOneEvolutionaryOptimizer.html


"Parametric estimate of intensity inhomogeneities applied to MRI" Martin
Styner, G. Gerig, Christian Brechbuehler, Gabor Szekely, IEEE
TRANSACTIONS ON MEDICAL IMAGING; 19(3), pp. 153-165, 2000,
(http://www.cs.unc.edu/~styner/docs/tmi00.pdf)


"Evaluation of 2D/3D bias correction with 1+1ES-optimization" Martin
Styner, Prof. Dr. G. Gerig (IKT, BIWI, ETH Zuerich), TR-197
(http://www.cs.unc.edu/~styner/docs/StynerTR97.pdf)




    Regards,


       Luis




BTW:  it is "Evolutionary", ... not "Revolutionary"   :-)


---------------------------------
Li, George (NIH/NCI) wrote:

> Hi Luis:
> 
> I have tested OnePlusOneEvolutionaryOptimizer, using the same set
> of dicom series with initial shifts in rotation and translation.
> The recorded iteration looks strange to me, however. It has many
> repeats with the same versor and vector for the transformation,
> as well as the same metric value. FYI, Mattes' MI metric is used.
> 
> Could you please take a look at the text file? And please let me 
> know if it is supposed to be or not?
> 
> Thanks for your advice.
> 
> George
>