Imaging informatics fellows build bridges in radiology

July 2, 2003

At many academic institutions and medical facilities, radiology informatics is still a nebulous entity that must address both clinical and technology needs. Teaching institutions across the nation are finding various ways to train physicians to meet the

At many academic institutions and medical facilities, radiology informatics is still a nebulous entity that must address both clinical and technology needs. Teaching institutions across the nation are finding various ways to train physicians to meet the clinical and technical requirements of a modern radiology department.

Three radiology informatics fellows from three different fellowship programs share their experiences and shed light on the action-packed and, at times, murky world of radiology informatics.

Dr. Arnon Makori Northwestern University

The imaging informatics fellowship at Northwestern is a one-year combined clinical and research program hosted by the radiology department and Northwestern Memorial Hospital. The program's design allows fellows to spend 60% of their time learning theoretical and practical aspects of PACS and other radiology information systems.

I was a full team member in the section's imaging informatics research laboratory, where I contributed to a variety of research projects. Able to choose from a multitude of possible projects, I focused on topics that interested me. During the year, I attended a number of professional scientific meetings where I was encouraged to participate and present. It is important for fellows to develop solid scientific research skills, especially if they intend to pursue an academic career.

During the clinical portion of my fellowship, I actively participated in the body imaging readout sessions and call duties. I was able to experience "life in the trenches" using the PACS at Northwestern and giving invaluable feedback to the PACS team. The fellowship gives students the ability to act as an important two-way conduit between the clinical radiologists and the PACS developers.

Participating in the body imaging fellowship allowed me to deliver presentations to fellows, residents, and medical students. Educational exchange greatly benefits other house staff, exposing them to informatics topics of which they may not have been aware.

While my fellowship ran fairly smoothly, there were a few bumps in the road. Trying to comprehend and master the extensive scope of the imaging informatics curriculum provided one of the major challenges of the program. While fellows in other radiology fellowship programs have already gained most of their professional knowledge during their residency, this isn't the case with the imaging informatics fellowship.

To become an active contributing member of the imaging informatics section, you must rely on a self-taught informal knowledge base. The knowledge base covers many nonmedical topics, including networking, database management, display systems, communication protocols, and more. Informatics fellows must promptly and meticulously pick up any missing knowledge. Initiatives are under way to formulate basic guidelines for an imaging informatics fellowship syllabus, and I strongly believe this will help prospective fellows in their preparation.

Another individual challenge was moving from a film-based department where I completed my residency to a 95% filmless state-of-the-art radiology department. I can now better understand the difficulties radiologists are facing when making the digital "jump."

The discipline of radiology is starting to recognize the importance and uniqueness of the imaging informatics specialty. An example is the RSNA's newly established sponsorship program of an institutional fellowship in radiologic informatics, commencing next year. I believe that professional appointments in imaging informatics, which now exist in only a few institutions, will become a widespread norm.

One of my goals is to advise and educate my professional colleagues. Today's radiologists must be technologically knowledgeable and enlightened to lead the discipline. They must participate in all levels of information system use: application design, purchase negotiation, and deployment of technology.

With my training, I see myself bridging the gap between the clinical and technological worlds and advising the industry on design and implementation of new information technology. Many nonmedical companies with brilliant information management solutions are unaware of their potential clinical applications. I see myself trying to identify these companies and helping them to find their complementary clinical niches.

Dr. Thomas Warfel University of Pittsburgh Medical Center

Diagnostic radiology is about more than just acquiring the images and producing reports; it's about answering the clinical questions. Every imaging study we perform is driven by a clinical management question in the mind of another doctor. Radiology informatics provides radiologists with relevant clinical data in addition to the images themselves. It also covers the interpretation chain, logging to whom "preliminary reports" were given, by whom, and when; distributing the final report; and explicating how any discrepancies between the preliminary and final report were relayed. Radiology informatics can increase diagnostic accuracy, provide clinically relevant results more quickly, and result in improved patient care.

I developed a love of imaging and image processing while an undergraduate in electrical engineering at Washington University in St. Louis. I was fortunate to become involved in PET imaging research at the medical center, and I also had the opportunity to work several summers at GE Aircraft Engines in Cincinnati, writing image processing software for an industrial CT scanner.

I wanted more direct involvement with medical applications and chose to pursue radiology, but at the same time I felt I needed a deeper technical understanding. I spent the next nine years in a National Institutes of Health-sponsored medical scientist training program, working on a Ph.D. in electrical and computer engineering from Carnegie Mellon and an M.D. from the University of Pittsburgh, followed by another five-year residency.

My fellowship at the University of Pittsburgh Medical Center was to be split: 50% clinical thoracic imaging and 50% radiology informatics. I wanted to maintain some clinically active focus, in part to retain the ability to identify relevant clinical problems, but also because radiology is fun. With chest CT on the edge of a technological explosion, I decided that thoracic imaging would be the best area in which to spend my clinical time. With regard to radiology informatics, my intent was to spend several months on a small-scale medical report authoring project and then get involved with existing research in medical image processing of low-dose, thin-section chest CT scans.

The year has not been easy. When serving two masters, neither is truly satisfied. Early attempts to split the day 50/50 failed. People in the reading room felt I was dumping work on them by leaving "early" if I used the afternoons for informatics, but I was "skipping the bulk of the work"

if I attended the reading room in the afternoons after the clinical morning rush was over. Attempts at alternating two-day/three-day and three-day/two-day splits have worked somewhat better, but they still lead to occasional over- or understaffing of the reading room.

As a further complication, a lung cancer screening study was set up this year, and my services were volunteered to create a miniPACS to allow storage and retrieval of anonymized medical images. The promised PACS infrastructure never arrived, and what was to be a 10% bite from my informatics time has grown to average 10 hours per week, usually at the expense of evenings and weekends, as the transitional system meant to bridge the gap has been pushed into regular use by the study.

Two full-time engineers were hired to extend the medical reporting system work under my direction. While the engineers offload technical work from me, there are new administrative issues to deal with. More of my time is spent in administrative meetings as the scope of the project widens. I have also experienced a learning curve in distinguishing between specifying system requirements versus implementation when communicating with those who do the implementing. Translating clinical needs into technical requirements continues to take considerable time, and my planned involvement with CT image processing research has been squeezed out.

Clinically, it has been difficult to generate momentum this year. Being part-time in the reading room meant I was automatically missing 50% of the interesting cases. Further, finding time for clinical reading continues to be a major challenge. Much of the autumn was spent preparing to retake two sections of the oral boards, then rushing to prepare a demo and a refresher course for the RSNA meeting, while maintaining the cancer screening miniPACS infrastructure.

A radiology informatics fellow requires not only clinical radiology and technical skills, but political skills as well. I know I haven't learned as much about thoracic imaging as I would have in a full-time fellowship. Still, I know more now than I knew last June, and I'm now able to hold intelligent conversations with the pulmonologists and thoracic surgeons here at the university.

Technically, I've gained a greater understanding of how speech recognition systems work and ways they can be improved to provide better performance. Moreover, I've come to understand that a transcriptionist provides much more than a voice-to-text service; punctuation, correction, and report formatting are difficult problems for a machine to perform. I've learned about report structuring as well as ways of representing that structure for machine understanding.

Politically, I've learned about the internal struggles for information control within and between the health system data centers, as well as the never-ending quest for funding. The ability to communicate with and persuade the nonclinical people who control the clinical data is every bit as important as the ability to solder and code.

I'm still frustrated by the vast numbers of things that could be done to provide better healthcare and by the even larger amount of inertia that needs to be overcome to achieve them. Many of the real problems are not technical; talented and affordable engineers are available. The real problem is the combination of politics within each medical establishment with a pervasive cultural complacency intrinsic to medicine. I would not be surprised if many of the true informatics innovations that occur in the next five years happen in smaller hospitals and clinics before they occur at the larger medical centers. While the larger centers may offer more resources, a person with a good idea and the passion to make it real may be able to bring it about more quickly at a smaller institution with less institutional inertia.

Dr. David Hirschorn Massachusetts General Hospital

My fellowship year has been the largest professional growth spurt I have ever experienced. MGH radiology provides a constant source of challenges and projects, and I get to pick the ones I want to tackle.

I had completed setting up an Israeli teleradiology project during the last few weeks of my residency, so it was no surprise that one of the first projects I undertook involved teleradiology to another international destination. This time it was India, where I dealt with virtual private networks, firewalls, and low-bandwidth connections, not to mention difficult people.

My largest project this year involved designing MGH's next batch of workstations. We formerly purchased workstation hardware and software from our PACS vendor. In contrast, we had always bought software only from our speech recognition vendor; they knew we didn't need to buy our PCs from them. Now that our PACS workstations are PCs, we arranged for our PACS vendor to sell us software only and load it onto PCs of our own design. That gave us the option of eliminating the second computer and combining speech recognition and PACS on a single computer. Can a single computer handle both applications? We're not sure yet, but we intend to find out.

Since high-quality flat panels are so cheap, and Windows 2000 makes adding monitors a piece of cake, we are exploring the utility of the five-monitor workstation. This project has taught me about the features and limitations of various graphics cards, as well as the nuts and bolts of monitor calibration.

My fellowship has also afforded me the opportunity to set up a series of lunches at which various vendors, mostly from the storage industry, demonstrate how their products can be used effectively in radiology. By observing the conversations between our senior computer engineer and the sales and technical representatives at those meetings, I learned the defining characteristics of storage needs for radiology systems, including RIS, PACS, and speech recognition.

Along the way, I have had several speaking opportunities to sharpen my lecturing skills, including a Harvard CME course in PACS. I also manage to engage in some clinical training in musculoskeletal MRI, but it's getting more difficult as time goes on this year. The most difficult part of my fellowship is indeed what my supervisor, Dr. Keith Dreyer, told me it would be: choice management.

I will return to clinical work part-time in July when I assume my new post at Staten Island University Hospital in New York City. I will be the director of informatics, overseeing the implementation of a new PACS, RIS, and speech recognition system. While I will be closer to my home in New Jersey, I am still thrilled to be remaining on staff part-time at MGH to oversee some of the projects I have begun and to publish their outcomes.

In a way, it's the best of both worlds. I became involved in radiology informatics early on in residency because it resonated with me, and the Society for Computer Applications in Radiology did a phenomenal job to help nurture that interest. But I wondered where it would take me: academia, private practice, or industry? In the end, my first job is somewhere between private practice and academia, and it seems like a great balance for me.

Related Content:

News