[Insight-developers] Re: ICP (Iterative Closest Point) Patent

[Luis Ibanez]

I agree with Stephen,

ICP is so basic, that it must be in the toolkit, but still
we want to make sure that it is clearly identified as a
Patented algorithm.


Having the set of directories that Stephen suggest in
point (1) seems to be enough for managing this code.


So far we will have to move:


   A) The implemnetation of marching cubes
          "BinaryMask3DMeshSource"
http://www.itk.org/Insight/Doxygen/html/classitk_1_1BinaryMask3DMeshSource.html



   B) For ICP, the fun part is that it is built
      out of pieces from the registration framework.

      In principle we just need to move Julien's
      implementation of the IterativeClosestPointmetric,
http://www.itk.org/Insight/Doxygen/html/classitk_1_1IterativeClosestPointMetric.html
      and move the three ICP Examples from

           Insight/Examples/Registration

      to

           Insight/Examples/Patented



It will be good too to create an additional Doxygen
group for Patented code, so they appear identified
in the documentation.


A Bug entry #1211 has been created for this issue:

  http://www.itk.org/Bug/bug.php?op=show&bugid=1211&pos=0

--


   Luis



------------------------
Stephen R. Aylward wrote:

> Time for the itkPatented directory...
> 
> A full solution would be:
> 
>     1) Create
>         Insight/Code/Patented
>         Insight/Testing/Code/Patented
>         Insight/Examples/Patented
> 
>     2) Since people can only patent algorithms, it seems like we
>         would not need directories below patented.
> 
>     3) InsightApplications would not use patented methods.
> 
> However, this leaves us with Lydia's original question...do we want to 
> maintain someone else's patented code???
> 
> So, option 2.  A partial solution:
> 
>     1) Create
>         Insight/Patented
> 
> This code might become out-of-date.   The user would have to manually 
> include the appropriate directories, etc - we could provide a sample 
> CMakeLists.txt file that should how to include Insight/Patented stuff in 
> a compilation...
> 
> I would hate to start deleting code from ITK.
> 
> Also, isn't Jay Udupa's group contributing Chris Taylor's patented 
> method.   There will probably be others too...
> 
> Stephen
> 
> 
> Luis Ibanez wrote:
> 
>>
>> Hi Raghu,
>>
>> Thanks for searching on the patent information
>> about the ICP algorithm.
>>
>> I'm afraid that serious advice on this matter could
>> only be provided by an intellectual property attorney.
>>
>> After reading the patent claim, it seems to me that even
>> though ICP is not the central aspect of the claim, the
>> method is clearly depicted in it.
>>
>> The worrisome part is that ICP is very widely used in
>> the medical field, in particular in image guided surgery.
>> If General Motors Corp. is actually enforcing the patent,
>> there would be many medical applications out there that
>> will be touched.
>>
>> The natural consequence is that people will move to use
>> a variant of the method, making it different enough for
>> not being covered by the patent claim. The question of
>> course is "how different" will be "different enough", and
>> that's the type of questions that only an Intellectual
>> Property Attorney could answer.
>>
>>
>>
>>   Regards,
>>
>>
>>
>>     Luis
>>
>>
>>
>>
>> ------------------------
>> Raghu Venkatram wrote:
>>
>>> Hi Luis,
>>>
>> ....
>>
>>> Also luis, I did a patent search on the Iterative Closest Point
>>> algorithm, as I remember Paul Besl, the primary author, mentioning in
>>> one of his talks that he ahd a patent.
>>>
>>> http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&p=1&u=/netahtml/search-bool.html&r=2&f=G&l=50&co1=AND&d=ptxt&s1=besl.INZZ.&OS=IN/besl&RS=IN/besl 
>>>
>>>
>>> Our application just like ITK is open source, please let me know if we
>>> would be violating
>>> any patents by using ICP for image registration.
>>> Thanks in advance.
>>> Raghu
>>>
>>>
>>
>>
>>
>>
>