By Dan Marcus
XNAT has been developed and maintained by the Neuroinformatics Research Group (NRG) at Washington University since its inception nearly 20 years ago. Much has changed during those two decades—cloud computing has emerged as the dominant IT environment; globally distributed software teams are made possible by Zoom, Slack, Github and dozens of other technologies; AI and related computational methods have pushed the limits of data availability and computing capacity. It’s a phenomenal time to be building informatics systems, but also increasingly difficult to accomplish in the context of an academic research lab, which just isn’t the right habitat for developing enterprise software.
We are therefore thrilled to announce that XNAT has moved to Flywheel, where it will continue to be developed as free open source software. Flywheel is a venture-backed biomedical software company focused on management of complex data in clinical and life science research. The Flywheel Enterprise platform has very similar DNA to XNAT, having been originally developed at the Center for Cognitive and Neurobiological Imaging at Stanford University.
This latest announcement builds on Flywheel’s commitment to XNAT, which began with its acquisition of Radiologics in the fall of 2021. Radiologics spun out of the NRG in 2010 to provide commercial support and features for XNAT. With the acquisition, all 15 Radiologics staff members moved to Flywheel and have since been supporting ongoing XNAT development, related engineering and science projects, and the legacy and new customers.
Flywheel currently employs over 100 brilliant engineers, scientists, regulatory specialists, technology leaders, and the sales, marketing, finance and other supporting staff needed to create a long-term stable enterprise software shop. It’s an exceptional environment in which to continue building XNAT and providing support to its global community.
The Move is Already Having an Impact.
The most noticeable impact of the Flywheel partnership is the unified development effort within Flywheel, which includes the folks who moved from Radiologics, new hires since the merger, and now the core development team that has moved from the NRG to Flywheel. We now have the largest XNAT team ever, including developers, designers, testers, and support personnel. This team leverages the strengths of veteran XNAT developers, some of whom have been developing XNAT code for over a decade, and newer members with unique and diverse technical backgrounds. Bringing the XNAT development effort into a single organization has allowed us to build more effectively, more efficiently, and at a higher velocity.
As a result, impactful new features have made their way into recent releases and many more are in active development. We’re particularly excited about a new module that will allow users to create and publish custom forms to capture novel data in structured, user-friendly interfaces. We also continue to collaborate with many external groups. Working with the NRG, a Jupyter notebook integration is currently in beta, and with the Institute for Cancer Research in the UK, new features for advanced image annotation and visualization have made their way into the XNAT viewer.
The unified team also enables more agile exchange of capabilities with the commercial products built on XNAT at Flywheel, including Flywheel’s flagship Real World Data (RWD) platform – Flywheel Discovery. RWD is increasingly essential to the conduct of large scale clinical, translational, and life science research, and Discovery provides the industry’s most advanced capabilities for creating imaging-based RWD. Building on XNAT is allowing Flywheel to move fast in this new space while contributing improved performance and scalability to the open source core.
The Future is Even More Exciting.
There’s more good news! The primary NIH grant supporting core XNAT development was awarded this September to Flywheel, providing funding through May 2026. The grant will take XNAT through 16 years of continuous NIH support, a remarkable stretch in today’s funding climate made possible by the exceptional XNAT developers that have shepherded it for many years now and the large and growing community of contributors and users around the globe. We anticipate using these funds to modernize XNAT’s core components and continuing to provide support to the XNAT community.
We are also thrilled to announce that Flywheel will soon be launching Flywheel Exchange, a secure data federation platform for engaging in collaborative research. It includes open data sets, private collaboration, and engagement with industry in data partnerships. The Exchange technology is being built to fully support integration with XNAT, and XNAT sites are invited to join this global network of researchers. We are in fact building Exchange to support an initial NCI-funded project that will create a data federation to enable the development of unbiased, generalizable AI for breast cancer imaging. It’s just the first of what we anticipate will be many collaborative research networks. Exchange itself will be rolling out in the coming months, and we encourage anyone interested in getting connected to Exchange to reach out to us.
The Community is Everything.
The recent XNAT Workshop was Flywheel’s first opportunity to meet with the community. Ten Flywheelios (yes, we call ourselves that) joined the weeklong event, many of us presenting in the sessions and all of us getting to spend a lot of time with the other attendees. We left the event humbled by the energy, insight, and commitment of the community and absolutely stoked to make XNAT as exceptional as the science our users are doing on the platform.
It’s also worth noting that a substantial NRG contingent joined the workshop and a considerable XNAT development effort continues in the NRG, with a focus on oncological imaging and preclinical imaging under NCI’s innovative Information Technology for Cancer Research (ITCR) program. Much of the Jupyter Notebook service is being developed in the NRG along with some very cool neuro-oncology, stroke, and brain connectivity workflows. The group also continues to push XNAT to be the AI platform we all think it needs to be.
With all these good vibes, perhaps you’re feeling the spirit to get more engaged with the XNAT community. Here are a few suggestions on how to get started. Definitely check out the sessions from the recent workshop (Videos of the talks are imminent!). The Jupyter functionality is a particularly straightforward way to code new functionality on XNAT, so once you’ve gotten through the basics, head straight to Andy’s and Hakim’s talks. And by the way… the XNAT/Jupyter integration beta has been released—consider it an early Christmas present from the XNAT team. Consider also engaging in a course on XNAT Academy. Beginning in the new year, Flywheel will be hosting regular virtual developer meetings, which will be a great way to get involved in the XNAT conversation. Finally, you can engage directly with the XNAT team and many others in the community on our Google discussion group, sign up to receive updates from us, or find Dan at the end of the bar at Duffy’s, where treating him to a Makers Old Fashioned will definitely get that feature you’ve been wanting in XNAT forever out in the next release.