Neuromorphometrics is building a database of expert-labeled MRI brain scans and we want to release it in a way that maximizes the common good.  Pure Open Sourcerers would have us do so for free, as in both gratis and libre.  I believe that charging for access and use of the database will be faster and more effective at creating more good.  I would love to hear what the NITRC community thinks.

Localizing and delineating neuroanatomy in 3D medical images is so tedious and requires so much expertise that few are willing and able to do it.  Automated methods are improving, but they always need correcting.  Creating high-quality results is so expensive that hardly anybody is willing to pay for it.

We provide access to our database as a subscription with a shockingly small fee for academic use and at less than cost for commercial use.  This funds the improvement of the database by allowing us to add more labeled scans and to parcellate finer regions.  The idea is to spread the cost over all who use the data.  The small academic fee allows nearly open access, while commercial users can afford higher fees because they stand to generate revenue from the data.

We could probably get away with using the open source model of openly releasing the data and then charging for services on the side, but a trickle of income provides for only a trickle of development.  Doesn't it make sense that we'll be able to improve and grow the database faster with our fee-based subscription idea?

Thanks for any thoughts!

Andy.