[Insight-users] Re: MultiResMIRegistration Example

[Luis Ibanez]

Hi Christina,

It seems that two lines are missing from your
parameter file.

The first four lines are about the fixed image:

- filename
- endianess
- number of pixels along each dimension
- spacing along each dimension

The next four lines should have equivalent
information for the moving image.

Then there should be two lines specifing
how to permute dimension and how to flip
dimension.

if you don't need to permute nor flip, these
two lines shoud look like:

    0   1   2
    0   0   0


After these two lines you can continue with
the number of level in the multiresolution
pyramid (that seems to be 6 in your file).
Then the subsampling rates... and so on.

Please try adding the permute and flip lines
in the rows 9th and 10th of your parameter file.

Also at the end of the file, after the set
of 6 learning rates to be used for the optimizer,
you should put the scale to be used for the
translation parameters.

In your file, there is a string at this position,
followed by two zeros. This doesn't fit the
expected format of the file.

---

You may want to follow any of the example
parameter files available in the subdirectory:

Insight/Examples/MultiResMIRegistration/SampleInputs

They are called:

          PracPETToPD.txt
          PracCTtoT1.txt


The Readme.pdf file describes in detail the
content of the parameter file.

Please let us know if you encounter further
problems using the example.


   Thanks


     Luis


=======================================



Christina Lee wrote:

> Hi Luis,
> 
> How are you?
> 
> I have been testing the MultiResMIRegistration example
> on a couple of MR images (they are actually images of
> the liver that have been previously segmented by hand)
> & I have gotten some really strange outputs. Please
> refer to the parameter file (reg1.param) for the
> parameters that I used. Please also refer to the two
> jpeg files. In mr_reg_result1.jpg, the liver on the
> left comes from the target volume whereas the liver on
> the right comes from the reference volume. I simply
> added the two volumes in mr_reg_result2.jpg & you can
> see that there isn't any overlap between the two
> volumes. Do you have any idea why I got these outputs?
> 
> Thanks for your help!
> 
> Christina
> 
> 
> 
> 
> __________________________________________________
> Do You Yahoo!?
> Yahoo! Finance - Get real-time stock quotes
> http://finance.yahoo.com
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
> 
> 
> ------------------------------------------------------------------------
> 
> mr2_pre_masked.raw
>        0
>      512     512      52
>      0.664062     0.664062      7.00000
> mr2_hepatic_masked.raw
>        0
>      512     512      52
>      0.664062     0.664062      7.00000
>        6
>        8       8       1
>        8       8       1
>     2500    2500    2500    2500    2500    2500
>   0.000320000 6.40000e-005 1.28000e-005 2.56000e-006 5.12000e-007 1.02400e-007
>      320     320     320     320     320     320
> mr2_hepatic_masked.reg
>        0
>        0
> dummy
> 
>