From ghosts to old town roads: An MRI odyssey

Many people think that open science is incompatible with IP. With this essay I try to challenge that dichotomy. Bear with me, it is counterintuitive.

In a couple of hours I will be giving a talk in Singapore at the annual meeting of the International Society for Magnetic Resonance in Medicine. The topic is the compatibility of open science and intellectual property (IP), something that is very near and dear to my heart. Yet, it took me until this morning to actually put my thoughts together in a coherent fashion. Call it procrastination, I call it last-minute inspiration.

Let’s start with a song:

This is 34 Ghosts by Nine Inch Nails, the industrial rock band that has surprisingly found their greatest recognition in the country charts¹. A sample from the above song formed the basis of Old Town Road, the record-breaking country rap by Lil Nas X that topped the US charts for 19 weeks in 2018!

So how is Old Town Road compatible with Ghosts? Well, Trent Reznor, the NIN frontman decided to release their 2008 album under a creative commons license (CC-BY-NC-SA), which allowed the song to be remixed and reused for free, as long as it was not used for commercial purposes. Ten years later, producer YoungKio sampled it, Lil Nax X discovered it, and they made a ton of money from it. But Trent Reznor was not left out. While their financial arrangement remains undisclosed, he is listed as Old Town Road’s songwriter and producer, making him an unlikely winner of a country music award.

Let’s try to bring this back to MRI.

In 2010 José Marques and colleagues published a paper on MP2RAGE, a popular MRI pulse sequence that has seen wide use for imaging at 7 Tesla. The accompanying code was released under a GPL-3.0 license, which made it possible for anybody to recycle the code, including Siemens. MP2RAGE is currently a product sequence on Siemens scanners, making José one of the most cited MRI scientists worldwide.

Because the sequence is in the public domain, it is now possible to remix it and bring it to other platforms. And that is exactly what we did a month ago, developing an MP2RAGE sequence using the vendor-neutral open-source pulseq framework developed by colleagues from Freiburg University. The development was done at Juntendo University in Tokyo, with collaborators from industry and academia that made it possible to do the work in less than a month.

In keeping with the music theme of this post, we called our project “MP2RAGE against the machine: Scanning in the name of reproducibility.” To the best of our knowledge, this is the first MP2RAGE implementation that runs on GE scanners, and other vendors are in the pipeline. Our next step is to deploy MP2RAGE on the proprietary RTHawk platform, developed by the company Vista. RTHawk is an FDA approved product that is closed-source, yet Vista has made it possible for customers to deploy open-source sequences on it. That is what we did in 2022 with our award-winning VENUS article, and MP2RAGE will be the obvious next step. So we have gone from an open-source sequence, to a proprietary Siemens sequence, to a vendor-neutral open-source sequence, to a a vendor-neutral open-source sequence on the proprietary RTHawk platform. The interoperability is evident. Open science and IP can, do, and should co-exist.

So far I have steered clear of the financial ramifications of open-source technology. One need look no further than Red Hat Linux and bitcoin to realize that open-source projects can make money. In 2019 we wrote an editorial that addressed this issue and explained how the very word “patent” implies openness:

Most would agree that transparency in science is an essential goal, but there are differing opinions on how open our science really should be. The questions surrounding open science and research transparency are particularly relevant for fields like magnetic resonance, in which much of the innovation over the past 50 years has been driven by intellectual property, including patents and copyrighted software, and there is a tight link between industry and academia.
…
There is a common viewpoint that open science is necessarily incompatible with the protection of intellectual property. However, even the etymology of the word “patent” implies openness, and historically patents have offered an important alternative to companies hoarding trade secrets.

As a matter of fact, in my native (Macedonian) language the word “patent” means “zipper”. Once a patent is published, the invention is unzipped, making it possible for other people to use it. The conditions for use are not always straight-forward, but in general patented technologies can be used for research as long as they are not for-profit. And if somebody does decide to profit from a patented technology, they can always sign a licensing agreement and pay royalties to the inventor. While not always used in their intended way, by definition patents are open-source technologies.

If this sounds counterinuitive, you should read my favorite essay on IP and open-source, called The Cathedral and the Bazaar. It is written by Eric S. Raymond, and it describes the differences between the Unix operating system (a cathedral model controlled by a centralized entity) and Linux (a bazaar model where everybody can contribute and modify code). In it, the unlikely champion of the bazaar model is (drumroll) MATLAB:

There have been other software products with a two-level architecture and a two- tier user community that combined a cathedral-mode core and a bazaar-mode toolbox. One such is MATLAB, a commercial data-analysis and visualization tool. Users of MATLAB and other products with a similar structure invariably report that the action, the ferment, the innovation mostly takes place in the open part of the tool where a large and varied community can tinker with it.

So next time somebody tells you that MATLAB is outdated, remind them that there are good reasons why it is still the most used language in the MRI community.

IP and open-science have coexisted for decades, and I hope that the MRI community will realize the potential of the bazaar approach to innovation. For that goal we have developed the idea of MRI FAIR, a vendor-neutral MR bazaar (app store) where developers can share their techniques with vendors and customers. This idea won the ISMRM Shark Tank competition in 2023, and we are now working together with other vendor-neutral projects (Pulseq and gammaSTAR) to create standardized implementations of MRI pulse sequences. We have decided to start with a unified MP2RAGE sequence, because its history makes it a perfect “34 Ghost” candidate. We will continue to remix, innovate and work with unzipped IP to create the Old Town Road of MRI.

P.S. A lot of the ideas for this essay fermented during yesterday’s MRathon, a community-driven effort to make MRI accessible, harmonized and vendor-neutral. Big thank you to everybody who joined us for the third hackathon for MRI professionals.

MRathon: Creating ghosts since 2019

1. Their song Hurt was covered by country legend Johnny Cash in this heartbreaking video directed by Mark Romanek