[Insight-users] INSIGHT JOURNAL: Test of Scattered Data Interpolation with Multilevel B-Splines provided by Nick

[Nicholas Tustison]

Hi Ni Dong,

It's quite possible that the code changed slightly from the version I  
used to create those test images which was at least a couple of years  
ago.  As you can see, the code has been modified several times since  
then.  Could you send me your test input and corresponding output  
privately?  I'll make a comparison and try to determine what accounts  
for the difference.

Thanks,
Nick


On Jul 9, 2008, at 12:34 AM, Ni Dong wrote:

> Dear Luis and Nick Tustison,
>
> I have posted a review to Insight Journal. Thanks for Luis's  
> reminding.
>
> In the meanwhile, I think there is some difference between the test  
> and the original result. I wonder if Nick did some postprocessing on  
> the result, such as intensity remapping, which is possibly the  
> reason of the difference. Could Nick give some suggestions about the  
> difference? Thanks firstly.
>
> In my opition, image registraion is one of the most important topic  
> of ITK. It is more preferable to generate a new class like  
> kernaltranform based on this B-Spline interpolation algorithm.
>
> I plan to apply this algorithm to deformable registration based on  
> two sets of correlated landmarks. I will post my results and codes a  
> little later.
>
> Thanks again for your attention.
>
> Best Regards
> Ni Dong
> Computer Science and Engineering Dept., The Chinese University of  
> Hong Kong
> ----- Original Message ----- From: "Luis Ibanez" <luis.ibanez at kitware.com 
> >
> To: "Ni Dong" <dni at cse.cuhk.edu.hk>
> Cc: <insight-users at itk.org>
> Sent: Wednesday, July 09, 2008 12:43 AM
> Subject: INSIGHT JOURNAL: Test of Scattered Data Interpolation with  
> Multilevel B-Splines provided by Nick
>
>
>>
>>
>> Hi Ni
>>
>>
>>            Thanks a lot for posting this review !
>>
>>
>> This is *exactly* what we hope readers will do with *every single*
>> paper in the INSIGHT JOURNAL:
>>
>>
>>  1) Download the code
>>  2) Build it
>>  3) Attempt to reproduce the paper
>>  4) Use the code with their own images
>>  5) Modify and improve the code
>>  6) Report back to the community
>>
>>
>> It is great to see a correct usage of the INSIGHT JOURNAL in action.
>>
>>
>> You may find interesting to know that Nick's class was recently added
>> to the Code/Review directory of ITK, and that it will be included in
>> the release of ITK 3.8. (note that given that the class will be in  
>> the
>> Review directory, its API may still change during releases).
>>
>>
>>
>> 1) One question:
>>
>>  Did you consider that the differences between your results
>>  and the ones posted in the paper were significant ?
>>
>>  Can they be attributed simply to computation precision ?
>>  and be disregarded ?
>>
>>  or do you think that something needs to be adjusted in the class ?
>>
>>
>>
>> 2) One Humble Request:
>>
>>  Could you please copy/past the text of your email as a review
>>  to Nick's paper in the Insight Journal ?
>>
>>  It will be great to have this text available for the readers
>>  who take a look at that paper.
>>
>>
>>
>> ...and, of course, Thanks to Nick Tustison for submitting a paper
>> that lends itself to *real reproducibility verification*.
>>
>>
>>
>>
>> Thanks !
>>
>>
>>
>>     Luis
>>
>>
>>
>> ----------------
>> Ni Dong wrote:
>>> Dear all,
>>>
>>> I have downloaded the code from http://hdl.handle.net/1926/140  
>>> provided
>>> by Nick Tustison . The codes were compiled using VC7. Test1  
>>> example(also
>>> given by Nick) was executed, which was used to interpolate the 2D  
>>> scalar
>>> images.
>>>
>>> The input image is "OriginalT2.png" also given by Nick. I just  
>>> modified
>>> the codes a little.  In the original program, the output image is
>>> "float" type. I just rescaled the gray from 0 to 255 and modified  
>>> the
>>> output image as "unsigned char" type.
>>>
>>> The result is a little different from the one given by Nick.  
>>> Attached
>>> please find the test result. The original data and  result can be  
>>> found
>>> from http://hdl.handle.net/1926/140
>>>
>>> The parameters used in my test are as follows:
>>>
>>> Splineorder = 4;
>>> Initial Numverofcontropoints = 5*5;
>>> Numberoflevels = 10;
>>> which is same with the ones presented in Nick's paper.
>>>
>>> Thank you very much.
>>>
>>> Best Regards
>>> Ni Dong
>>> Computer Science and Engineering Dept., The Chinese University of  
>>> Hong Kong
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> Insight-users mailing list
>>> Insight-users at itk.org
>>> http://www.itk.org/mailman/listinfo/insight-users
>