Thanks Bennett.

Let me justify the scientific case a little bit, also in response to Matthew's very relevant comparison with scientific software.

First, the raw data are freely available (OASIS for now I believe), so anyone who wants to segment these and make them freely available can do so. Apparently no one else so far has. Apparently no one has been asking, at least successfully, NIH for money to do this either. Personally, I'd love for this to happen, but then, apparently so far no luck.

Regarding the analogy with scientific software - Matthew is right of course, but I see one crucial difference: consistency.

To me, one of the major benefits of what Andy's company is doing is that all scans are segmented via the same well-specified, tested, and consistent procedure. This, to me, looks like something a community-based effort my have trouble to replicate. Having 60 segmented brain scans is great, but it's even greater if they're all segmented into the same regions and with comparable reliability.

Would I rather see these data produced by an NIH-funded project and shared freely? You bet... but until that happens, I am happy we have an alternative available.

By the way - one thing Andy hasn't mentioned: you can contract them to segment data for you. So say you have a funded project and want your data segmented, you can pay them to do it. And here's the kicker - if I remember correctly, you can pay them a larger chunk of money, and then you can distribute the resulting segmentations as freely as you wish.

TR