A pioneer in imaging informatics writes on behalf of an open source approach in radiological informatics.

In the mid ’90s, I became interested in the Linux operating system. Some of the physicists down the hall were using it in their lab. At that time I was running the Solaris OS from Sun on a Sun SPARC 20 computer. I had also purchased a 160 MHz PC and loaded Linux on as the OS. It performed at about the level of the SPARC 20, a more expensive machine with a proprietary OS. Being cheap, I decided to switch to Linux and also decided to switch my SPARC computers to Linux. A company called Red Hat distributed a version of Linux for the SPARC 20. However, when I tried to load the Linux system on the SPARC 20, it crashed and would not proceed. To solve this, I posted a help note on the Internet using the USENEX mail forwarding system…it ran around the world. The answer I received was local, from someone in my building a few floors up. As it turned out, he knew of the problem and the solution: the needed modified Linux OS was in a far-off place in Eastern Europe. We hooked up my SPARC computer over the Internet and brought up a foreign modified version of Linux. It worked!

I could never have achieved the same results in the proprietary world. The lesson learned was that the Internet lets people cooperate and share and provides almost instantaneous help on free software. Was I happy. I have been using Linux since then. It is much less expensive and more reliable than the Microsoft world. It is also less expensive than the other UNIX systems available in the proprietary world. Also in the early ’90s, I was selected as the chairman of the Radiological Society of North America (RSNA) Electronic Communications Committee (ECC). The RSNA let this committee freewheel within reason and supported it at a very healthy level of funding. That is when we on the ECC, in contact with the American College of Radiology/National Equipment Manufacturers Association committee, decided that Digital Imaging and Communications in Medicine (DICOM) was a good idea. We convinced the upper level powers of the RSNA to initiate support of DICOM development and created an RFP for programming help, which Washington University in St Louis won. We also decided to make DICOM open source: the programs would be given away so as to establish the standard. This was a lesson learned with the Internet format TCP/IP. It was developed at Berkeley under Bill Joy and freely distributed, thus the success of the Internet and a single standard. DICOM too has become a standard, although with some minor hiccups along the way.

I do not think I need to repeat the history of Linux, but note it has relied on the cooperative interaction fostered by the Internet and it is now open source. Credit that to Linus Torvalds ( www.cs.helsinki.fi/u/torvalds/ ) along with the Free Software Foundation led by Richard Stallman ( www.gnu.org ).

There is a huge amount of open source software on the Internet. OpenOffice ( www.openoffice.org ) provides a standard for text document exchange and presentation exchange. It has gained a large following in that it can run on most popular operating systems and thus exchange documents between them while being free to use.

What is the value of open source software? First, the software is free for anyone to use and distribute in the original language in which it was written, making it available to change and easy documentation. Also, because the source is freely examinable, security of open source software is at a high level. There are literally thousands of eyes on the Internet examining each piece of open source. This also provides a set of experts who will reply almost immediately with solutions to problems, unlike the proprietary software world, where only a few people have an ability to solve problems and try to guarantee high levels of security…they cannot do it! Linux, an open software OS, has been shown to be much more reliable than many of the more entrenched products. Why? For all of the above reasons!

While Microsoft may be the prominent OS in the United States, Linux is on its way to becoming the world leader. There are a number of countries that have decided to use Linux as the official OS, including Germany. It should be noted that one major computer vendor has seen the future and has invested more than a billion dollars in Linux. And you can now buy PACS servers from one vendor that use the Linux OS. Where will the support for Linux come from? University computer labs use Linux as an OS because it is open, allowing the computer science students to study the OS in depth. In fact, many programmers now run a version of Linux at home on their personal computers.

So what does this have to do with radiology? We are already using open source in radiology through the Internet protocol, TCP/IP, and DICOM. There are fairly powerful products in our world and examples of where open source in radiology is going. There are groups developing open source PACS software and open source RIS software ( www.minoru-development.com/en/healthlinks.html ). The RSNA and Healthcare Information and Management Systems Society are working on an open source standard using DICOM and HL-7 to connect all the computers in a medical setting. The RSNA now has released an Open Source teaching file, although it is written in JAVA, a proprietary language supported and not released by SUN.

What does this mean to the future of radiology? Radiology is fast becoming a computer-centered specialty: the software we are using is becoming more complex and needs to adopt an open standard so that the world of radiology can progress. We are just at the beginning of this evolution, as computer-assisted diagnosis (CAD) along three-dimensional displays of information becomes central to our practice. MRI will soon produce studies of more than 10,000 slices and will require a computer to help the radiologist view and analyze the case. Think of the situation where the computer makes a diagnosis or modifies the image in a way that creates a wrong diagnosis. Should not the radiology department have the ability to understand those workings? And what might the lawyers say? Doctor, you made your wrong diagnosis with a black box you do not understand or control?

Is there then a business model that will support open source? The question is already answered by what the aforementioned computer vendor is doing with Linux. Novel just purchased SUSE, a Linux distributer. A radiology practice already supports software upkeep…would not that work with open source? And by using open source in CAD, the ability to analyze images could be shared and advanced at a much high rate of evolution.

Laurens V. Ackerman, MD, PhD, is director of medical informatics, Department of Radiology, Rush Presbyterian Hospital, Chicago.