Merlina Trevino

Work list management tools are finding their way into PACS offerings, but that hasn't stopped many radiology departments from launching efforts to develop their own solutions, according to studies presented at the RSNA meeting.

More and more institutions are moving their work lists away from a first-in, first-out model to a system that red-flags high-priority studies. This way, busy departments can zero in on the most urgent cases.

Radiologists at Cincinnati Children's Hospital Medical Center, for example, launched RadStream last summer. This Web-based, homegrown automated triage system ensures priority for the most acute cases and immediate communication of urgent results to referring physicians. Its use has improved reading times for inpatient and outpatient exams and reduced final sign-off times for emergency department exams, according to Dr. Mark Halsted, Cincinnati Children's chief of radiology informatics.

RadStream also provides documentation and keeps a permanent log of all communications regarding cases, including the time, date, and staff involved in each communication event. An in-house survey revealed that the hospital's radiologists like the new tool. After implementation, radiologists reported a 22% decrease in daily interruptions, which translates into nearly 1500 radiologist-hours saved per year, Halsted said.

"We need to help busy radiologists prioritize their tasks," said Dr. Matthew Morgan, a radiology resident at the University of Pittsburgh Medical Center.

Morgan and colleagues developed an in-house software algorithm that highlights time-critical exams and automatically prioritizes a radiologist's work list. When developing the system, researchers defined two types of urgent exams: those from the ER, and those read by radiology residents or by clinicians providing preliminary reads.

Morgan's system is different from traditional filtering methods, which look first at division and study time as parameters. The logic built into the software prioritizes exams by patient location as well as STAT and urgent notifications by physicians. Use of the system in the neuroradiology division has decreased the time from exam completion to final interpretation from 9.6 hours to 7.8 hours for urgent studies.

Addressing the need to streamline communication with referring physicians, Morgan's group developed four urgency levels:

Category 1. Findings are life-threatening, need high levels of attention, and require immediate synchronous communication. Minutes count in patient management.

Category 2. Findings need high levels of attention but can be managed in minutes to hours. They require reliable asynchronous communication such as e-mail.

Category 3. Findings can be treated before the sun sets. They can be communicated via synchronous/interrupted modes of communication.

Category 4. Findings do not warrant high levels of attention. Communication could be conducted through a radiology Web portal for physicians.

The software provides context-specific messaging for each different type of communication.

"One-size communication does not fit all," Morgan said.

Dr. Dan Cohen, a radiology resident at Massachusetts General Hospital, and colleagues developed a system to notify their requesting physicians of urgent findings. The alert is triggered by the radiologist and sent via e-mail to the referring physician. It does not contain the actual exam results but instead provides a link to them. To avoid desensitizing physicians to the alerts, they are not used for minor findings, Cohen said.

During an eight-month period, radiologists at MGH completed more than 390,000 exams and sent 8210 alerts. About 70% of these were viewed by physicians, but more than 20% of the alerts did not generate return receipts, indicating that physicians may not have opened or received them.

A survey revealed several reasons why physicians missed the alerts:

They had already seen the important finding in the HIS.

They had never ordered a particular study.

Their systems were outside the hospital's firewall.

These missed alerts could be reduced by hiring dedicated staff to follow up and make sure that physicians receive and respond to them, Cohen said.

Dr. Kevin McEnery, an associate professor of radiology at the University of Texas M.D. Anderson Cancer Center, continued the theme of what he called smart radiology.

"The radiology workflow needs to be aligned with clinical workflow," McEnery said.

He outlined how his institution moved away from the typical first-in, first-out mode of interpretation by developing a work list engine that gathers data from the PACS, the RIS, the EMR, and the HIS. The engine allows researchers to manage the workflow of close to 1500 cases. Radiologists in the department have expressed satisfaction with the workflow, McEnery said.

That satisfaction is due in part to eliminating repeated interruptions caused by ad hoc prioritization, according to McEnery. Inpatient studies and STAT cases receive higher priority.

These homegrown approaches reflect a dearth of work list solutions from the commercial community, said Dr. Paul Chang, chief of radiology informatics at the University of Pittsburgh, during a panel discussion. But expecting vendors to have a product that pleases everyone may be unrealistic, given the various idiosyncrasies of users.

One solution would be for vendors to provide service-oriented architectures, which would allow departments to customize products to fit individual needs, said Dr. Keith Dreyer, vice chair of radiology computing and information sciences at MGH.

Articles

Homegrown work list tools provide custom solutions

Security event monitors, an emerging technology in security software, can help protect radiology departments from the increasing onslaught of network attacks, according to researchers at the University of Pittsburgh Medical Center.

The monitors correlate data from various network system defenses and provide users with one unified source of security information, said Dr. Barton F. Branstetter IV, associate director of radiology informatics at Pittsburgh. Branstetter spoke at a SCAR University session on security.

The Health Insurance Portability and Accountability Act has made security one of the hottest topics in healthcare.

"It's one thing to lose a paper chart, but it's another to expose your medical records to the outside world," Dr. Paul Chang, director of radiology informatics at Pittsburgh, said during the session.

An emerging problem in network security is the growing frequency and complexity of attacks, according to Branstetter. New event monitoring tools can help hospitals gain a better understanding of the various ways their networks may be infiltrated.

Branstetter used a pyramid analogy to describe a hospital's network security. At the base of the pyramid are static defenses such as firewalls. These are essentially "dumb" defenses that do not analyze attack types or even recognize that they have been breached. They simply block certain transactions and log those results.

The second line of defense relies on analytical software tools, including intruder detection systems that "sniff" data and look for suspicious conversations such as broadcast messages and increased traffic. Other tools at this level of the pyramid are probes that mimic hackers and can alert administrators to weak points in the system, system integrity verifiers that identify changes in system files, and log file monitors that detect rejection event patterns.

But even these more intelligent security software systems can't deal with security data overload, he said. Some larger entities can be attacked by up to 10 MB of security data threats per minute. With so many assaults, often originating at a variety of sources, managing the amount of incoming data can be nearly impossible.

This is where security event monitors, which sit at the top of the security pyramid, come in. They gather data from all the systems below them and correlate the information by time and type of event.

These tools essentially provide a meta-analysis of all collected security data, helping administrators to determine the timing and sources of attacks. Thus armed, they can prepare for impending attacks, find system loopholes, and develop containment strategies, Branstetter said.

The technology is still maturing, and only a few products are on the market. The University of Pittsburgh uses many elements from the middle tier of the security pyramid but is waiting for security event monitoring technology to fully mature.

"In medicine, we tend to be way behind the curve in adoption of this sort of technology, compared with business and others in academia. It is important that we keep abreast of such developments," he said.

New software crystallizes network security big picture

Security event monitors, the latest advance in security software, can help defend radiology departments against the growing onslaught of network attacks, according to researchers at the University of Pittsburgh Medical Center.

The monitors correlate data from various network system defenses and provide users with one unified source of security information, said Dr. Barton F. Branstetter IV, associate director of radiology informatics at the center.

"An emerging problem in network security is not the type of attacks, but the growing frequency and complexity of those attacks. What makes it even worse is the growing complexity of the systems that need defending," Branstetter said at a SCAR University session Friday. 

New event monitoring tools can help hospitals gain a better understanding of the various ways their networks may be attacked. Branstetter used a pyramid scheme to describe a hospital's network security. At the base of the pyramid are static defenses, such as firewalls. These are essentially "dumb" defenses that do not analyze attack types or even know that they have been breached. They simply block certain transactions and log those results.

Moving up the pyramid, the second line of defense consists of more analytical software tools, including:

Intruder detection systems that "sniff" data and look for suspicious conversations such as broadcast messages and increased traffic 

Probes that mimic hackers and can alert administrators to weak points in the system 

System integrity verifiers that look for changes in system files 

Log file monitors that detect rejection event patterns. 

But even these more intelligent security software systems can't deal with security data overload, Branstetter said. Some larger entities can be attacked by up to 10 megabytes of security data a minute. With so many attacks, often coming from a variety of sources, managing the amount of incoming data can be nearly impossible.

This is where security event monitors, which sit at the top of the security pyramid, step in. They gather data from all the systems that sit below them in the security pyramid and correlate the information by time and type of event.

In essence, these tools provide a meta-analysis of all the collected security data. This analysis arms administrators with one information source that allows them to determine the timing and recurring sources of attacks. They can then prepare for impending attacks, find system loopholes, and develop containment strategies, Branstetter said.

New software class provides big picture on network security

Radiology departments need not fear irreversible compression of medical images, according to a Canadian review of the medical literature.

Two independent reviewers concluded that lossy compression is clinically acceptable for image compression, said Peter Bak, Ph.D., project director of diagnostic imaging architecture for Canada Health Infoway. The organization promotes adoption of electronic health information systems in Canada.

Some data loss does occur with lossy compression, which leads to concern that a misdiagnosis could result if relevant information was lost, he said Thursday at the SCAR meeting.

Bak and colleagues commissioned two independent researchers to systematically review the research on lossy compression. They scanned radiology and related journals, surveyed radiologists in the U.S. and Canada, and sampled professional radiology organizations in both countries for position statements on the use of irreversible image compression. The reviewers studied 120 papers, most dating from 1998.

They found that two main compression algorithm categories are currently in use: discrete cosine transform-based and wavelet-based algorithms.

The researchers also discovered that different modalities and anatomic structures have different compression tolerances; there is no one-size-fits-all compression scheme. Their consensus was that a 50:1 compression ratio is acceptable for large-matrix images, with 25:1 optimal. An acceptable ratio for small-matrix images is 20:1, with 10:1 optimal. 

Acceptable compression ratios varied between modalities, from 40:1 for MRI and up to 90:1 for nuclear medicine studies.

The adoption of electronic health records will require optimum use of networking capabilities among institutions, and compressing images can go a long way toward achieving that goal, according to Bak.

The Canadian Association of Radiologists has endorsed lossy compression for primary reads.

"We in Canada have a good motive to use lossy compression. We see lossy compression as having a positive impact on our national EHR plans if we can make it happen," he said.

Investigation of the barriers to and drivers of lossy compression adoption will require more than a research review, Bak said. In the future, more work will be done to determine legal risks, assess regulatory frameworks, determine economic impact, and provide clinical and performance evaluations.

Merlina Trevino

Articles

Homegrown work list tools provide custom solutions

New software crystallizes network security big picture

New software class provides big picture on network security

Lossy compression provides viable option for storing medical images