SCAR 2005 Educational Sessions
The annual meeting of the Society for Computer Applications in Radiology (SCAR) brought record numbers to Orlando this year from June 2nd-5th. Attendance jumped 18 percent to 2,975 at the meeting that marked SCAR's 25th anniversary. High attendance at the many educational sessions showed they were a core interest of attendees. Here is a summary of a sampling of educational sessions.

The Economics of Filmless Radiology

SCAR U analyzed the economics of filmless radiology with the help of Paul Nagy, PhD, director of informatics research at University of Maryland, who provided strategies for buying PACS on a small wallet. Eliot Siegel, MD, vice chair of radiology at University of Maryland, focused on the economic challenges of dealing with a PACS divorce. Bruce Reiner, MD, director of radiology research at the VA Maryland Health Care System, and Sean Casey, CEO of Virtual Radiologic Consultants, teamed up to share the concept of grid radiology.

PACS on a Small Wallet

The PACS market is evolving into a more mature market. Hardware prices are dropping, and software integration is becoming increasingly critical, Nagy said. Hardware accounts for less than 30 percent of the cost of PACS.

Commercial off-the-shelf hardware can meet many needs in radiology, according to Nagy. A commercial 2 megapixel monitor is suitable for reading almost all studies except DR and CR.

While hardware and software have relatively short lifespans, standards last 25 to 50 years. "Learn IHE and DICOM and use it where applicable to solve problems," Nagy said. Integration is a key concern; although all vendors conform to DICOM, the standard is flexible with varying levels of integration among products. Finding out whether or not proposed vendors connect at the RSNA Connectathon can save integration time and costs after the purchase.

Nagy offered four vendor selection tips:
  • Don't select one vendor prematurely.
  • Market competition is a positive; use it to your advantage.
  • Learn from all vendors.
  • Rely on the KLAS report for user satisfaction information.
Several other details can trim PACS costs. Line item pricing allows buyers to see and verify the cost of all components - particularly hardware that can be purchased from other suppliers. Asking for a software-only price ensures that hardware is not tied into legacy components. The contract should specify payment on acceptance, which can be measured by functionality or filmlessness.

A decision matrix that evaluates the total cost of ownership, functionality, technology, integration capabilities and industry relationships can guide the purchase decision, Nagy said. Relevant users should stick to relevant components. That is, radiologists should analyze functionality; and IT should review technology.

Economics of a PACS Divorce

Both first and second time PACS buyers can learn from PACS divorces, Siegel said.

Once a hospital or imaging center and PACS vendor decide to part ways, data must be migrated from the old PACS to the new one with data migration - taking approximately 30 percent as long as data acquisition. That is, if the first PACS operated for five years it will take 1.6 years to migrate the data to the new system. Costs for data migration range from 40 cents to $1.20 per study, Siegel said.

Data migration options include third- party migration vendors or the facility as most PACS vendors are unwilling to assume responsibility for the task. Siegel offered a combination approach as a possible solution to the data migration quandary. That is, maintaining the first PACS as the radiology PACS and replacing the enterprise wide distribution system, providing the new enterprise system with copies of new studies and enabling access to priors. "This may be the first step in the migration strategy or a long-term solution," Siegel noted.

Siegel concluded with advice for profiting from the spate of PACS divorces.
  • Include obsolescence protection in the contract.
  • Limit pre-paid service contracts.
  • Consider a software-only PACS migration if you have a strong and motivated IT department.
  • Focus less on the RFP and more on vendors' track record.
  • Don't forget to calculate time and cost required to swap vendors.
  • Don't plan for a long-term relationship with one vendor.

Grid radiology is the way of the future

Grid radiology is a concept of the radiology environment of the future that aims to solve problems of increasing image utilization and data complexity as the radiologist shortage grows, Reiner predicted.  

It applies the concept of grid computing, which links multiple computers from different locations as a unified source, to the radiology business to optimize the radiologist supply/demand imbalance. Potential forms could be a large enterprise teleradiology practice, loosely associated radiologists or independent groups working on the same platform or network or a giant website where radiologists could find work or hospital could locate radiologists.

Necessary infrastructure includes a meta-PACS that ties multiple PACS together as one system, IHE XDS to integrate multiple enterprises and a virtual radiology environment.

In addition to easing the radiologist shortage, grid radiology could decrease reimbursement, tie reimbursement to quality and increase sub-specialization, Reiner predicted.

Inside the reading room

Creating the ideal reading room was the focus of a multicomponent session by Paul G. Nagy, PhD, director of informatics research at University of Maryland addressed reading room design. Alan Hedge, PhD, director, human factors and ergonomics laboratory at Cornell University, focused on reading room ergonomics, and Kenneth C Johnson, MSc, of Kenneth Johnson and Associates presented architectural case studies.

Reading room design

Nagy presented reading room design experience at his former employer, Medical College of Wisconsin in Milwaukee as the site implemented digital radiology. Prior to implementing digital, Medical College of Wisconsin relied on a traditional ballroom design in its reading room. Problems with this configuration include light pollution, stray film, noise and clutter, which compromised alternators.

After defining local goals, such as maintaining access for clinicians and increasing productivity, the next step is to conduct research about how other sites approached design. Sites on the leading edge include UCLA and Baltimore VA Medical Center.

Nagy offered industry rules of thumb for the digital reading room:
  • Don't mix alternators and workstations. Place an alternator in a separate room.
  • Don't have workstations face each other.
  • Don't ignore ergonomics of the desk & chair.
  • Create space for consultation.
  • Don't ignore ambient and task lighting.
  • Take step to reduce noise pollution.
  • Don't limit workstations to one for every two alternators.
Finally, solicit radiologists' input on items like reading session length, mouse placement and headache incidence.

Medical College of Wisconsin eventually implemented three workstation pods with more private areas along the walls and one private reading office. To minimize transition time, the college scheduled the project over two 72-hour weekend shifts with radiologists temporarily relocated to a conference room. Ceiling drops for power, communications and the network facilitated the fast track project.


Hedge's session focused on defining ergonomics as a process, not a label. He also offered tips for implementing ergonomics into practice. Hedge says the ergonomic design process centers on seven questions and corresponding answers.
  1. How do you know if you have an ergonomic design? Check user sources like employee complaints and medical records, and conduct discomfort surveys. An ergonomic audit, either systematic observation of users or a work environment checklist, can provide additional data.
  2. How can you quantify injury risk factors? Postural targeting methods and subjective and objective measures can help determine injury risk.
  3. How can you develop ergonomic guidelines for the reading room? Specify environmental conditions, space and layout, work postures, furniture and equipment and work patterns and include the rationale for each in guidelines.
  4. How can you minimize injury risks and maximize comfort and performance?Decrease postures like wrist extension, ulnar deviation and perching that place users at risk.
  5. What are comfortable environmental conditions for the reading room? Apply standards for thermal air quality, lighting, noise, vibration and indoor air quality.
  6. How can you evaluate the impact of ergonomic design changes? Conduct ergonomics evaluation after any changes.
  7. How can you facilitate a participatory ergonomics process? Train users in work postures and provide them with appropriate tools to enable the ergonomic process.

Architectural case studies

Kenneth Johnson demonstrated an improved four-step process for reading room design, claiming the traditional process is flawed and costs time and money.

Sites should obtain computer aided design (CAD) and have all drawings sent in CAD, allowing users to get a feel for the proposed room. Use functional drawings to analyze workflow. Spend 1/2 a day or longer in sites that have implemented the desired outcome, and discuss design plans and issues with them. Simulate the design with all key stakeholders with a scale model or crude mockup. Johnson concluded, "The best configuration is a function of many factors. Well-designed space is a necessity, not a luxury."

Tips for buying RIS, optimizing EMR, adding evidence-based radiology

Establishing priorities for buying RIS, optimizing the EMR for radiology and evidence-based radiology were among the priorities discussed at the SCAR U session on IT presented by Janice Honeyman-Buck, PhD, of University of Florida who instructed participants on the ins and outs of buying or upgrading a RIS. Katherine P. Andriole, PhD, of Brigham and Women's Hospital provided suggestions geared toward optimizing the EMR for radiology. And Ramin Khorasani, MD, MPH, of Brigham and Women's Hospital focused on evidence-based radiology.

"Step one," Honeyman-Buck said, "is to specify what you want." Hospitals and imaging centers should form a task group composed of key users including IT and billing staff, radiology administration, technologists, radiologists, film librarians, patient registration staff and the PACS administrator.

The task group needs to write specific goals to include in the RFP. "Use the exam workflow to look at processes," recommended Honeyman-Buck. She offered sample goals and objectives:
  • Integrate and automate front office and billing processes with electronic claims submission to accelerate cash flow and reduce lost charges
  • Interface with PACS to integrate images and reports
  • Increase referring physician satisfaction
  • Process more patients with the same FTE count
Objectives should be measurable, and the RFP should include all requirements such as order entry, ICD9 support and inventory management. Sites upgrading a RIS should list needed features and specify migration of current data and when and who completes the migration. The RFP should detail expected performance - such as a query for the patient record should return results in less than two seconds.

Honeyman-Buck concluded with 10 mistakes to avoid when buying a RIS:
  • Not using a structured process
  • Not defining needs
  • Hiring a consultant with a bias
  • Too much attention to bells and whistles
  • Not including key users in the selection process
  • Buying more than needed
  • Allowing the vendor to drive the process
  • Letting the 'powers that be' choose the system
  • Confusing the salesperson with the product
  • Not using the RFP process
Andriole outlined the functionality and technology necessary to optimize the EMR for radiology. The EMR is a multi-faceted system that includes the HIS, CPOE, RIS, PACS, report generation systems, decision support and teaching files.

Key issues in EMR integration include standards and data models such as DICOM, IHE and XML; availability of single sign-on and a master patient index and context and user sensitivity.

A variety of factors can predict the success of the EMR, Andriole pointed out. Ease of integration into the workflow is a primary issue; however, sites may re-engineer workflow to optimize the system. Does the EMR add value? Is the user-interface simple? Are there well-defined goals and metrics of improvement? Is the system configurable, fault-tolerant and redundant? Are there organizational support and a vision for the project?

Evidence-based radiology begins with a few requirements, said Khorasani. Sites must be prepared to close the knowledge gap and accept that physicians make mistakes; systems are needed to prevent mistakes, he said. Knowledge must be distilled into a brief, actionable format accessible at the point of care in clinical context. Thus, it should provide brief summaries of research relevant to the specified indication and test.

Evidence-based radiology is needed in a number of areas. This includes determining if the right test is ordered, if the exam is performed and interpreted correctly and if the right action follows the test. Multiple studies of inpatient and outpatient exams across modalities reveal that 10 to 25 percent of radiology exams are unnecessary and inappropriate.

Khorasani offered a list of barriers to evidence-based radiology including leadership, information systems, knowledge gaps and costs. Intelligent CPOE that incorporates context-specific decision support is a necessary infrastructure for evidence based radiology. Productivity gains can provide an incentive for physicians to use CPOE, so integrate it into the electronic record, Khorasani said.

At Brigham and Women's, 85 percent of referring physicians are ordering electronically and evidence-based radiology decreases the number of inappropriate studies.    

Security is not so boring after all

Security may be a boring topic that few in healthcare want to think about, but it's of vital importance to patients. Oh, and real security doesn't just come out of a box. Those were some of the key points of a presentation by Paul J. Chang, MD, of the University of Pittsburgh Medical Center called Introduction to Information System Security.

Chang offered a number of ways in which organizations should evaluate their security practices:
  • Building a Security Framework - Emphasize procedural changes that boost security and focus on physically controlling access to data and planning for power outages.

  • Securing the Network - Firewalls must be constantly reconfigured, encryption-based data transmission strategies (VPN, SSL) must be used, security-software must be regularly updated, and 'back door' network access must be monitored to secure your network.

  • Securing Computers - Mandate the use of anti-virus software with mandatory regular updating, desktops in public areas should be locked down, and laptop use should especially be monitored for viruses and other harmful software.

  • Securing Users - All users should be required to have authentication before accessing the network, guest accounts should be banned, administrative rights should not be given to general users, user lists should be audited regularly, users should be trained in security practices, and HR policies should add reinforcement.

  • Security for Wireless Technology - A three pronged approach should be used to secure wireless technology, including authentication which should require User IDs with designated rights, and mutual authentication from both ends of the wireless connection with external authentication from a server such as RADIUS or LDAP; encryption must be used to safely encode data; and message integrity can be assured by utilizing software that provides code that evaluates the content of a message for tampering.

Image alignment impacts image interpretation

A presentation by Bradley J. Erickson, MD, PhD, of the Mayo Clinic in Rochester, Minn., featured details of a SCAR-supported R&D project that evaluated the possible benefits of comparing and matching prior images to more recent studies using a variety of techniques.

The study involved the recruitment of 20 sequential and comparable CT and MR examinations including neurological and body studies. The initial review registered the old with the new, but the second review included only the latest exam. The time differential in coming to a conclusion about the image in both cases was measured.

One of the key findings was that the body images did not benefit from the comparisons because of the disparity between the older and more recent studies. However, comparisons of older and newer neuro exams did see some promising results with study registration a key factor, though not in all cases.

Erickson emphasized that adopting of standard scanning protocols is key to fully realizing the promise of such study comparisons, especially in the case of body studies.

PDAs and other powerful 'toys'

Richard H. Wiggins, III, MD, University of Utah School of Medicine provided a lighthearted and interesting overview of handheld computing technologies in PDAs and Other Personal Technologies. He subtitled the presentation: "favorite geek toys."

Wiggins ran through a series of PDA (personal digital assistant) products and their key features. Starting with the huge line of PALM PDAs, he then drilled through a large assortment of alternatives including Mobile Windows PDAs, Psion PDAs, Linux, and finally the Blackberry which was highlighted for its ability to grant instant access to email around the clock. All of this information came with the warning that some of these devices are as expensive as full PCs.

Wiggins then dispensed with PDAs and offered an exuberant glance at a number of personal technologies such as the interesting (though scary) wi-fi sniffer - which allow a user to snoop around within unsecured wireless environments, iPods, USB drives, remove mouse pointers, and an assortment of others.

The development of these personal technologies has led to PANs (personal area networks), computer networks for connecting a series of devices within a small space so that they can communicate either through a wired or wireless environment. Bluetooth technology, which uses radio waves and can transmit data at a very slow rate between different devices, can be used to connect a PAN, which is then called a Piconet, said Wiggins.

CT & MRI get new DICOM data sets to calm volume overload

After more than a decade of use and with the immense growth in the volume of CT and MRI studies, it's time for a new DICOM standard - which was outlined by Bradley Erickson, MD, PhD, of Mayo Clinic in Rochester, and David Clunie, MBBS, chief technology officer of RadPharm.

"Important data elements like diffusion, perfusion and new CT scanning modes aren't standard across vendors. Nor do images process or display well across all vendor workstations," Erickson said. Furthermore, the series numbers PACS use to hang studies do not necessarily reflect practice.

Enhanced DICOM objects address the problem by including many more standard mandatory attributes and terms and by allowing multiple frames in a single object. "This enables greater interoperability, more effective hanging protocols and less dependence on private attributes," Clunie said. The richer set of mandatory descriptive attributes will facilitate downstream display of images on various workstations.

New objects include the presence or absence of contrast to make it easy to distinguish contrast use. Mandatory coded anatomic regions and laterality will improve hanging protocols.

New objects support advanced applications, facilitating visualization, temporal changes and quantitative and analytic applications. The addition of color by functional paradigm facilitates specific applications like functional MRI (fMRI).

The new version includes raw data classes to store images in PACS, so that they can be regurgitated for reconstruction later. And DICOM now supports storage of spectroscopy data.

A final problem addressed by the new DICOM is the immense network burden of new multislice CT scanners. New multi-frame objects don't change common headers; information isn't replicated and latency between storage requests is reduced.