INTRODUCTION

Imaging informatics is a powerful field representing the intersection of physicians, physicists, and clinical staff with computer scientists and information technology (IT) professionals from across a diverse range of academic, clinical, and corporate settings. The field is challenged to continuously identify, develop, embrace, and promote innovations that have profound effects on the ways in which health care is delivered. The myriad types of interactions, the computational intensity, and the wide variety of applications required for successful imaging informatics practice call for clearly defined and easily implemented standards.

The process of sharing source code and programs has played an important role in the development of the field of imaging informatics. Open source development has been critical in accelerating the adoption of the digital imaging and communication in medicine (DICOM)1 standard, the de facto global IT standard for medical imaging.2 Open source applications have become synergistic partners of open standards by providing a powerful means of building reference implementations.3–7 Some of these applications are of great practical value to picture archiving and communication systems (PACS) administrators in diagnosing integration problems commonly seen in clinical settings.

Open source is an umbrella term—at once a noun and adjective—that describes a development method that allows researchers to exchange algorithms and IT professionals to share tools.8 This process not only facilitates individual advances but, by pooling resources and freely publishing improvements, results in more rapid consensus development and validation of new tools. Open source communities have been able to utilize the Internet to identify and engage contributors from around the world to effectively distribute efforts to develop advanced and robust information systems. As imaging informatics matures, open source is likely to play an increasingly important role in both basic imaging research and clinical diagnostic and therapeutic practice.

An open source program is one in which the source code to the software application is freely available for distribution along with the program itself. Access to the source code gives the knowledgeable user the ability to create “fixes” to the program and make modifications that may be of benefit either in specific circumstances or for general use. More than 190,000 programs are currently hosted by the two major open source repositories at http://sourceforge.net and http://freshmeat.org.

Several of the most successful open source programs in diagnostic imaging are profiled in this special edition of the Journal of Digital Imaging. One rationale for this special edition is the growing divide between the technical and end-user perspectives in the medical imaging community. Although the topic of open source development might seem more suited to IT professionals, it is important that all members of our community be at least aware of the potential significance of these powerful tools for continued advancements.

For some imaging department professionals, knowledge about open source is essential. Open source programs and utilities should be essential parts of the PACS administrator’s armamentarium for diagnosing problems and maintaining a smoothly running PACS operation. Several platforms are available to enable even those developers without advanced skills to innovate in specific imaging research applications. For any medical professional interested in understanding the growing complexities of DICOM, one of the best approaches is to download and start using open source programs to see exactly what makes DICOM tick.

The central tenet of open source is the free and collaborative exchange of ideas that combine to enhance innovation. Although sometimes depicted as counter to traditional models of proprietary software development, open source has proven to be a successful approach for selling services based on sequentially enhanced versions of the freely available programs. Many commercial vendors in radiology now participate enthusiastically in open source efforts. The distributed software methodology of successful open source applications has also resulted in high quality, as many developers peer review the contributed code. This has been demonstrated to be superior in some cases to commercial software applications.9,10 This article is intended to provide the reader with an introduction to the ways in which open source has been used in diagnostic imaging and to provide as a ready reference a list of programs that are open source and available for working with DICOM.

THE VALUE OF OPEN SOURCE IN ROUTINE IMAGING

PACS administrators should consider using open source tools for several reasons. Programs in which the users participate as developers are most often powerfully intuitive. Open source development methodology shortcuts the traditional barrier in development with degrees of separation between the users and developers of an application.11 This observation is reinforced by the fact that systems administration tools that are open source are usually quite competitive with their commercial counterparts.

Another benefit of the open source approach is in circumventing traditional bureaucratic and budgetary restraints. Depending on the specific practice or institution, the process of purchasing new IT tools may be cumbersome and time consuming. Moreover, budgets can be trimmed at the last minute, effectively overturning well-laid planning for new systems. Familiarity with open source resources and the ability to easily integrate these tools not only provides “no-cost” programs, it frees the PACS administrator (in at least this one area) from dependence on the timelined purchase approval process. These benefits extend to the other end of the traditional purchase process: open source tools are available immediately (no ordering or delivery time required), an advantage that can help to avoid slowdowns or hold-ups, for example, in tightly scheduled modality integration projects.

Open source also carries distinct learning curve advantages. The cost of new software applications (not insignificant investments) pales beside the time costs incurred in bringing users up to speed and proficiency. Familiarity with one open source program accelerates the learning curve on all its subsequent iterations. Moreover, open source tools are portable. They can move with the skilled PACS administrator from job silo to job silo within or even across institutions, without the need to negotiate with employers for the purchase of the software platform with which that administrator is most experienced.

Predicting the utility of an untried software tool is not a foolproof science. Not every tool lives up to expectation or advertising. With open source programs, one seldom needs to go all the way back to the metaphorical drawing board to find a solution or a different tool, and the process does not require either begging for budgetary forgiveness or asking for additional funding.

Finally, an open source community offers an excellent avenue of growth as an imaging professional, even for individuals whose principal interests are not in software development. The collaborative exchange of ideas in an open forum not only helps to “grow” skills on an as-needed basis, but provides a virtual setting in which accumulated expertise results in recognition for some participants as international leaders and experts in their areas.

FROM INNOVATION TO ACCEPTANCE

Before DICOM, one vendor’s imaging system—and all of its components—had no way of communicating with other vendors’ systems. Each hospital purchased a complete system from a single vendor, with work stations, archive, modalities, and film printers all on an isolated proprietary network. The Radiological Society of North America commissioned two groups to develop DICOM communications tools in the early 1990s as a means to accelerate the adoption of the DICOM standard.

The OFFIS group (Oldenburg, Germany) developed DCMTK, a collection of libraries and applications implementing large parts of the DICOM standard. It includes software for analyzing, constructing, and converting DICOM image files, handling offline media, sending and receiving images over a network connection, and other features.

The Electronic Radiology Lab at the Mallinckrodt Institute of Radiology (St. Louis, MO) originally developed the central test node (CTN) software to support cooperative demonstrations by medical imaging vendors and has been upgraded as imaging technologies and requirements have advanced.

Vendors were allowed to take the source code from both the DCMTK and CTN efforts and implement these in their clinical systems to support the standard. Vendors also found that the source code was particularly useful in understanding how DICOM was intended to be implemented and the challenges that would likely be faced in everyday operations. These efforts were successful and were key factors in transforming the medical imaging industry from an ad hoc approach to today’s best-of-breed environment. Both DCMTK and CTN remain available today and are used by clinical vendors to implement the DICOM components of their software.

Many software vendors who first perceived open source as a threat to their proprietary offerings have now come to appreciate several advantages that open source can leverage for their companies’ continued success. The medical imaging industry has partnered with the open source community in several ways, including but not limited to:

  • Reducing support costs: Costs associated with support and maintenance of older code can be a substantial drain on commercial vendors. Open sourcing lowers the support costs of legacy applications and allows commercial developers to focus on adding value instead of patching older software.12

  • Reducing development costs: Open source software can lower the cost of development by using existing open source programs to add to the features of a vendor’s product. This is especially useful in efforts to ensure that new products conform to DICOM and HL7 standards and requirements.

  • Adding business advantages: By releasing programs as open source, a commercial vendor can attract talented developers whose efforts mean that the programs will be more widely used. The company can then sell service and support on these same open source programs.

OPEN SOURCE AS A COMMUNITY EFFORT

The true mark of success of an open source project is seen when it matures into a community. Few open source projects evolve to this level, but the process of evolution itself is an example of collaborative quality control. Internet-based evolution in an open source community has been characterized as a winnowing process. For every 10 people who use an open source program, only one person is likely to submit bug reports or feature requests. For every 10 people who submit bugs, only one person is likely to either submit a fix along with the report or submit a new addition.13 This winnowing process also tends to provide positive results. Eric S. Raymond, one of the pioneers of open source, is often quoted as saying that “many eyes make bugs shallow.”14 Because of the participatory and highly focused nature of collaborative development, community-based open source projects can rival and, in many cases, be superior to commercial code.

Many community-based open source projects also provide a substantial level of customer support through forums and through iterative, shared documentation on Wikis. Users find answers to common and not-so-common questions by searching the forums for reports from others who have encountered and solved similar problems. One telling metric of the merit of an open source project is the level of activity on its online forums. Vibrant and active forums provide responsive customer support from enthusiastic volunteers. Another way to judge the viability of a program is to determine when the last update was made to the software. Good open source projects update their applications monthly, weekly, or, in some cases, daily. Another key to assessment is that the most vibrant communities are those that include not only developers but nonprogrammers. This mix helps to ensure that the project is not a “developer-only” tool, but one with broader utility and appeal.

The copyright on open source projects is sometimes wryly referred to as a “copyleft” because it serves to remove more restrictions than it imposes on the use of the software. Open source projects are typically based on one of two types of licensing models. The first is derived from the Berkeley Software Distribution (BSD)15 licensing family of permissive free software licenses, which provide the language and structure needed to allow users to redistribute software and even sell it commercially. The CTN and DCMTK DICOM implementations were released under a BSD-like license, helping the effort to accelerate DICOM adoption. The second derivation is based on the GNU General Public License,16 which allows free use of the software but requires that any modifications be contributed back to the public domain.

EVALUATION CRITERIA FOR OPEN SOURCE PROJECTS

The biggest challenge for the user who wants to get started in open source—even in the relatively circumscribed area of medical imaging—is in knowing where to begin. Once a potentially useful project has been identified, the next challenge is in determining whether it is a fully featured program or merely an abandoned collection of code. In our large urban imaging department, we use the following preliminary evaluation criteria when assessing an open source project:

  • Web site appearance and documentation. Good documentation is perhaps the best indicator of a truly successful open source project. Paradoxically, documentation is often the last task that developers address. In the best open source efforts, documentation is written not by the core developers but by individuals with the points of view of users. Documentation begins with the front page of the project Web site, which should be easy to navigate and informative. Mature projects will include screenshots, installation guides, user guides, and separate guides for developers.

  • Activity and utilization. The second best indicator of a successful project can be found in the statistics provided by open source repositories like http://Sourceforge.net and http://Freshmeat.net. These sites provide relative activity metrics that indicate how often the project is downloaded, when the developers last updated the code, the number of registered users subscribed to the project, and information about activity on bulletin boards.

  • Ease of installation. Although many open source applications are supported on a range of platform operating systems, this does not guarantee that applications will be easy to install or will operate as plug-and-plays. Installation failures are most often the result of a lack of documentation or inadequate validation tests on a multitude of hardware platforms. Immature open source projects may lack installation instructions, include multiple dependencies that are not packaged together, or require specific versioning of those dependencies. If background investigation, as evidenced by reports of other users, indicates that a program is quite difficult to install, this may be a sign that the project has not matured sufficiently to be a good investment of the user’s time.

  • Technical support forums. A busy and active support forum for a program is not a sign of a buggy application but of the existence of a large group of enthusiastic users who are helping each other get the most value out of the application. In a truly vital open source community, response times for answers to even the most difficult questions can be quite short. The existence of a frequently updated section on common questions (FAQs) is another sign of an active and successful project community.

OPEN SOURCE PROGRAMS IN DIAGNOSTIC IMAGING

Many freeware viewers have been described elsewhere in the literature17–19 and are not included in the listing presented here. Many open source projects are likely to be missing from this compilation because no central repository or even a common categorization methodology is available for open source projects. Tables 1, 2, 3, 4, 5 and 6 are divided into categories of DICOM servers, DICOM tools, DICOM viewers, teaching file systems, research oriented tools, and Web-based PACS systems.

Table 1. DICOM Servers
Full size table
Table 2. DICOM Tools
Full size table
Table 3. DICOM Viewers
Full size table
Table 4. Research Oriented Tools in Image Processing
Full size table
Table 5. Teaching Files
Full size table
Table 6. Web-based PACS
Full size table

CONCLUSION

Open source plays an essential role in the field of imaging informatics, facilitating the spread of beneficial innovations into utilization and practice. The list of projects in imaging informatics summarized here shows the scope, depth, and energy of community innovation in our field. The Internet has enabled an international community of distributed development, connecting disparate professionals in the goal of advancing the state of medicine through IT. Open source tools are pervasive in our field and should be an integral part of every PACS administrator’s troubleshooting kit.

The best and most continuously rewarding developments in modern IT seem to draw on community-based efforts. The Society of Imaging Informatics in Medicine and the Journal of Digital Imaging have their shared roots in the oldest grassroots-based medical imaging effort: the Radiology Information System Consortium. The open source community is a remarkably rich resource from which imaging professionals across the spectrum of practice and research can derive both the everyday tools and the synergistic strategies that will guide us through a future of rapidly changing technologies and demands.