[Insight-developers] Patented code

[Stephen R. Aylward]

run away...run away.... :)

I think it is a volation if you infringe on any single claim.  Including 
derivations that would be easily contrived of by anyone "skilled in the 
art."

My vote is to create a patented directory, put the active shape stuff in 
it, and only keep such patented code if we have a letter from the patent 
holder.

I am going to write a draft policy for the ISC...short and simple, 
ideally...  Copied from other open-source efforts...

s



Miller, James V (Research) wrote:
> One question that we need to answer is what does the patent "cover"?
> 
> 
> The patent uses language like
> 	
> 	sensor to measure points on a surface
> 
> and, 
> 
> 	mathematical model of an object
> 
> and, 
> 
> 	mean square error metric
> 
> Most of medical image devices do not measure points on a surface. At best, 
> we use gradient information to infer a surface. Does grouping a CT image
> and a gradient operator make it equivalent to a "sensor that measure
> points on a surface"?
> 
> What about registering centerlines of vasculature?  The centerlines are
> surely
> not on any surface. Centerlines are presumably "inside" a swept cylinder.
> 
> Does registering two images based on features of those images in an ICP 
> framework violate the patent? The patent refers to registering points
> to a mathematical model.  Does it cover image to image registration?
> 
> Does the patent cover ICP algorithms that do not use a mean square error
> metric?
> 
> I can't recall whether you have to violate all the claims of the patent to
> be
> infringing or whether violating any one of the claims triggers an
> infringement.
> 
> For instance, the patent appears to cover a "shape inspection" application, 
> where data from a range sensor or CMM is compared to a geometric model. If
> our
> application is not shape inspection, do we violate the patent?
> 
> 
> 
> 
> -----Original Message-----
> From: Stephen R. Aylward [mailto:aylward at unc.edu]
> Sent: Wednesday, September 29, 2004 2:58 PM
> To: Insight-developers (E-mail)
> Subject: [Insight-developers] Patented code
> 
> 
> I just spoke with a patent lawyer.
> 
> By distributing patented methods as code we are technically commiting a 
> crime called "contributory infringement."   Even if we don't use the 
> code ourselves, we are making it easier for others to do so.   No amount 
> of documentation or disclaimer insulates us.  Being "open source" does 
> not insulate us.   If someone notifies us that some code is patented, as 
> someone just did regarding ICP, and we do nothing, then additional laws 
> are being broken.
> 
> We must have written permission from the patent holder to distribute any 
> code that implements a method covered by a patent.  In the event that 
> the patent holder has granted an exclusive licenced of any portion or 
> application of the patented technology, then we also need to get written 
> permission from that licensee.
> 
> Seems like this is a huge headache.  As Bill has been telling us :)
> 
> It seems like we should not take on the burden of getting permission.
> 
> Option 1:
> Anyone who contributes code must attain permission.   That permission is 
> likely to have stipulations regarding the conditions of distribution 
> (patented directory, etc.).  Are we willing to adapt our 
> distribution/directory to fit requirements as they develop?
> 
> Option 2:
> We don't allow patented code.
> 
> Other options?  Comments?
> 
> Stephen
> 
>