Summary:
The authors describe a Wiki-based site they are developing that will serve two important purposes:

1) To inform the users of available medical image data on the web. 

2) To provide logistical support to establishing, maintaining, and using medical image databases.   

Open Science and Reproducibility:
Consider that conducting open-science and testing reproducibility are frequently hampered by an apparent lack of open-data.  Also consider that one of the key contributions of a successful open-source toolkit is often that it simply provides a centralized repository for effective algorithms.  It is in light of these two considerations that the authors of this paper are introducing the concept of open-data centralization.

Open-data centralization is achieved by dedicating a wiki page for each available medical image database.  Those wiki pages provide a standardized format for describing the content, contributors, licenses, modalities, etc of the referenced image database.

Furthermore, the website provides logistical tools.  These tools include things such as (1) suggested text for including in an IRB form so that collected data can be contributed to an open-database and (2) code for anonymizing dicom data.   These tools are often the most stressful part of establishing an image database.  The authors work in this area is a significant contribution to the field.

At this time, the wiki pages are still in development, but there is a critical need for this specific type of effort.

I personally look forward to contributing what I know to its success.

Suggestions for future work:
Provide effective (form-based) methods for searching the wiki pages and getting summary reports (e.g., number of MR scans with 1x1x1mm or better resolution of brain tumor cases available across all sites).  This will require further standardization and detailed reporting, but it is a worthwhile future goal.

Summary:
The authors describe a way for collecting web-pages that provide databases with free medical images for validation and testing. The method is to provide a Wiki where users contribute locations and description of databases. Their conclusion of gathering database information in a very simple and effective way seems to be promising.

Hypothesis:
Not Applicable. Hmmm. I guess their idea of using users to collect medical image databases could be called an hypothesis. If so IMHO it is an excellent idea.

Evidence:
Not Applicable. Hmmm. Again, if their evidence is the number of databases which are already listed and described on the web-page then I have to admit that there is not much evidence yet. The only database listed with a description is the NLM Mayo Imaging Collection, which unfortunately is off-line at the moment. But already the list of potential databases is very useful evidence - one can still google for them.

Open Science:
The ideas of open science are one-to-one represented with this contribution by listing open-source databases and the descriptions of their usage conditions and by letting the interested community participate.

Reproducibility:
Not Applicable.

Use of Open Source Software:
Not Applicable.

Open Source Contributions:
Not Applicable.

Code Quality:
Not Applicable.

Applicability to other problems:


Suggestions for future work:
I think it would be useful that the authors start to equip the Wiki with more descriptions of databases. How about the Insight data collection project? I think as soon as the repository has some more descriptions and links in it, people using it will get a bad conscience and try to contribute if they can. Another database that I found while I was working on medical images is the 'Internet Brain Segmentation Repository'. Since there are a lot of undescribed databases on midr.org, I'm not sure if it is already contained with a different name - nevertheless I'm going to add it - however to provide the description it would perhaps be a good idea to ask the maintainer of IBSR for a short entry in the Wiki.

Requests for additional information from authors:
All the needed information is given in the paper.

Additional Comments:
After thinking about this topic a little more - perhaps the authors should also write mails to all the maintainers of the different sites that they list in their Wiki. For them it wouldn't be much of an effort to write a small Wiki page about their repository and the midr Wiki would grow and get some critical mass to start being very useful for other researchers.

Overall: 

All in all I find this contribution very useful and will definitely follow its progress and try to contribute if possible.