The Open Anatomy Project is a research and educational project designed to make medical information stored as anatomy atlases widely available. It was developed at the Surgical Planning Lab in the Department of Radiology at Brigham and Women’s Hospital in Boston. It is based on the research activities of hundreds of medical researchers and scientists, as well as the open source developments of a large distributed community.
For most people, an anatomy atlas is an illustrated textbook of anatomy with labeled organs, systems, and structures. We broaden this definition to refer to any annotated spatial description of anatomical structures. In fact, atlases don’t need to be anatomical.
See our About page.
See What’s wrong with today’s atlases?
Our atlases are meant to be:
Free and open atlases make new discoveries and a broad social mission easier. It means everyone can use and everyone can create. Also, in many cases we have already been paid by the National Institutes of Health (NIH) to create this data. We’re giving back this investment by the American taxpayers to benefit the public and the world. Under the NIH’s open data mandate, our data (and the data from all other recipients of large NIH grants) is supposed to be made public. We believe we can so that in a way that is useful, that enables new scientific discoveries, and that serves the fundamental human need for medical knowledge.
Rather than rely on just one authority or publisher to create our atlases, we will enable the broad medical community (clinical and research) to publish their own information as anatomical atlases. That might mean highly curated atlases, subsets of existing atlases for specific purposes such as education, or individual case studies.
We are working other researchers in the community to develop an atlas data format that will provide a common foundation for representing many different kinds of atlases. Our atlases have 3D models, medical imaging data, and annotations, but other atlases could even be two dimensional drawings or annotated photos. (We need help refining this data format.)
Anyone will be able to use this format to represent their data. Anyone can publish their atlases on the web so that compatible atlas viewers can read them. Anyone can put their atlases in version control systems like git, though they might not be ideal for handling large models or image data.
On top of data representation and viewers, Open Anatomy wants to build communities around atlases the same way GitHub has build communities around software projects. GitHub provides centralization that enables discovery, discussion, and collaboration. We believe that having a single place to look for published anatomical information is important, and unavailable at the present time.
We believe that the combined expertise and resources of the broad medical community far exceeds that available through any single publisher or single medical authority.
We are focusing on free and open atlases. However, our technology will be open source. The same tools can be used for “closed” atlases. For example, a hospital might have a private instance of an atlas data hub to compare specific medical cases. We also anticipate allowing atlases to be developed in private before being released publicly.
The question of correctness is both practical and deeply philosophical. There is natural concern about the publication of incorrect anatomical information. The answer is the same as that to the question, “How can I trust an anatomy atlas such as Gray’s Anatomy?”: we know and trust the reputation of the publisher, we rely in experts (including ourselves) to verify its contents, and we can compare it to other atlases.
We envision a system where known entities such as physicians and medical schools can publish atlases. Just like GitHub, we can assess the reputation of the authors. Second, we can use experts to verify an atlas using community standards. Any expert will be able to provide public comments on an atlas. For atlases we endorse, we will have an expert review committee approve atlas content. And finally, we hope to have multiple atlases of different parts of the body, allowing direct comparison between atlases.
No, though it is important for the community to know the development state of an atlas. Distributed software development and the emergence of GitHub has shown that software can be developed incrementally by a community, where different members provide different skills.
Our vision is that some individuals and groups will have high-quality or even unique medical data that they could publish, others will have expert segmentation or classification or modeling software, others will be expert in labeling, and still others will be able to create a highly meaningful visual rendition of the data.
Each of these creators can publish their own work, building on the work of others.
Yes. We believe that Open Anatomy’s development model can bring the collaborative and incremental development characteristics of distributed software development to open data, where they have not been common.
In open software development, we have effective version control tools for text-based computer software. We have good models for “diffs” (differences between versions). We understand how to “fork”, or copy, a code base, and merge it in larger. We have “pull requests”: social communication built around suggestions for change. Collaborative software development is build on these kinds of tools.
In open data, especially images, these tools generally don’t exist. Finding the differences between two images is harder than differencing text because we need to understand the semantics of the images. Open data tends to be monolithic: all the data is stored together, so differencing is impossible, forking involved wholesale copying of very large data,and no mechanism exists to give back changes.
The Open Anatomy data model splits atlas data into different types of data, each with its own characteristics:
The Open Anatomy data design also includes a layer concept, where new atlases can use pieces from existing ones by reference rather than by copying. We believe these data characteristics will enhance the community aspects of open data by enabling incremental development and “give back” of improvements.
Absolutely not. We believe there is plenty of opportunities for for-profit atlases and open atlases to exist together. Publishers can provide rigorous editorial review. They benefit from the availability of free atlases. They can provide enhanced software tools for viewing or editing their atlases. They can link journal content into their atlases, using the atlas as a spatial index into their publications.
We would hope (and appreciate!) that publishers might provide some of their high quality materials to create open two- or three-dimensional open atlases. By doing so, they would provide a framework for additional developments and annotations by the community, which the publisher could then incorporate into paid content if they chose.
The Open Anatomy Project is at the proof of concept prototype stage. We have a draft (but incomplete) data format, a prototype-quality atlas viewer, and a collection of atlases. We have NIH funding to develop the data format to represent research atlases and to produce a robust atlas gallery. We have a design that needs more discussion and ideas. We have collaborators who are interested in donating atlases. And we have medical collaborators who desperately need high-quality anatomy teaching materials.
In short, Open Anatomy is suitable for evaluation and discussion, but not robust enough for rigorous use. In particular, the atlas viewer is not suitable for mobile or tablet use.
The Open Anatomy vision will only happen with community support. Here are some things we need:
Without hyperbole, we believe the Open Anatomy Project can change the world. We need help to make this vision happen. Please join us!