Duke Medicine’s Imaging Informatics Game Plan

Sept. 22, 2014
There’s no easy route to creating an enterprise imaging informatics strategy, and extensive planning is critical to a successful implementation. For insight into what such a plan entails, one can look to Durham, N.C.-based Duke Medicine, a three-hospital, 1,500-bed academic medical center.

There’s no easy route to creating an enterprise imaging informatics strategy, and extensive planning is critical to a successful implementation. For insight into what such a plan entails, one can look to Durham, N.C.-based Duke Medicine, a three-hospital, 1,500-bed academic medical center. In May 2012 Duke set about the task of selecting of an imaging IT vendor, as it embarked on creating an enterprise-wide imaging strategy and archive that could accommodate rapid imaging growth and meet the needs of every medical specialty. That’s a tall order that required extensive planning, according to Christopher Roth, M.D., associate chair of radiology for health information technology and clinical informatics and director of imaging IT strategy at Duke Medicine. In September, at a meeting of the Vendor Neutral Archive Institute in New York (sponsored by Milwaukee, Wis.-based TeraMedica), Roth provided a detailed account of his organization’s planning process.

Duke’s imaging strategy includes several goals—managing, retrieving and storing images as well as their care, exchange and interpretation. Storage is central to many of those goals, not only in terms of capabilities and functionalities, but also in terms of the different locations of images and a platform for their storage and management, he said.

The Case of an Enterprise-Wide Imaging Strategy

Roth observed that radiology and cardiology save roughly the same amount of data, and almost always have a picture archiving and communications system (PACS). Both departments might have several reasons to wanting to revise their storage architecture. Among them, they might want to get out of a PACS vendor lock-in, or they might have a distribution problem, where not all of their imaging customers can access images in the way they want. In addition, images that are used for teaching or research purposes must be anonymous, which changes the way they are stored, he said.

Enterprise imaging systems can help to improve workflows in areas of specialties that have not been integrated into an enterprise electronic health record, Roth said. One area that can benefit are in free-standing clinics—single radiography rooms or portable ultrasound devices—where images are often used in clinical decisions, but storage is often an afterthought.

Similarly, specialties such as obstetrics, gastroenterology, dermatology, and ophthalmology often have acquired PACS or developed imaging IT systems on their own, he said. At Duke, those specialties were set up locally, so that only a small number of people could view the images, even though they could be of interest to a wider audience, he said.

Duke handles volumes of images that are growing at a fast clip. While cardiology and radiology will always be a factor in imaging, Roth said demand is growing rapidly among other specialties as well. One area is pathology, which he expects to be a major area of imaging IT strategy in the future. Duke produces 1,000 histology slides and 200 cytology slides per day, and the volume is growing at 1 to 2 percent per year. In the future, he estimates, pathology images will account for 4 terabytes of data in a single day, not including whole specimen imaging or cytogenetics imaging. That’s roughly what Duke stores in cardiology and radiology imaging in four to six months today, he said.

Roth also identified video as an area of potentially rapid growth, driven by educational purposes as well as telemedicine. Speaking of the latter, Roth noted that doing synchronous conferencing of adequate length and quality requires significant space. At Duke, which runs a fairly large telemedicine program, a full neurological exam conducted remotely on a potential stroke patient encompasses cranial nerve, motor, speech, sensory, and broad cognitive exams. The amount of storage a provider organization keeps depends on medical considerations as well as the fact that payers are focusing more attention to telemedicine, and are requiring documentation of the video interactions, he said.

Who are the Biggest Consumers of Images?

Roth noted that there are many different groups of consumers in a healthcare system, and said it’s important to have an image management strategy that reflects that reality. “You want the architecture you use to store these images and distribute them across the enterprise strategically hits anything you want,” he said. “You want a flexible archive, something that is going to stand the test of time.”

He named five major image user groups.

  • Large volume creators: Cardiology and radiology use large volumes of data in terms of exams and required storage, and typically own the workflows in creating the images.
  • Passive image consumers: Primary care, family practice, pediatrics and internal medicine are at the other end of the spectrum. They generally do not have deep imaging knowledge.
  • Small volume creators: “Back office” imagers who walk the patient from the exam room and the radiology room for an x-ray or ultrasound. Specialties include ophthalmology, gastroenterology, or the ED. This group is important to an imaging archive, because clinical decisions are made off the images, yet many times the images are not managed as optimally or as securely as possible.
  • Power users:  This is a small group of sophisticated sub-specialist users—typically vascular surgeons or neurosurgeons—who are interested in post-processing images. Often these are secondary captures that are stored somewhere so they can be used for when they take a patient in the OR. Those captures generally can’t be put back in the exam where they came from.
  • Patients: This is a group that should be engaged as savvy image consumers. In his view, consumers have increasing interest in high-dollar decisions, especially as payment models change. The imaging archive plays into patient engagement for the enterprise. Patients are increasingly interested in seeking out reports and images through the patient portal, and the archive needs to be able to link up with the portal, as well as allow patients to easily move images between hospitals. Long-term storage of images needs to be ensured to comply with state and federal requirements for breast imaging and pediatrics.

Roth said that organizations with silos of radiology, cardiology, dermatology and other specialties should consider IT economies of scale with respect to simplifying interfaces, support and maintenance. He added that an enterprise image management strategy should integrate multiple image viewers, by providing a universal viewer for any type of DICOM or non-DICOM still image or video; a PACS viewer with a higher level of functionality; and advanced visualization applications such as profusion imaging and three-dimensional reconstruction. “There are a lot of people interested in doing imaging at a medical center, and once you start opening up that functionality, it becomes very interesting to doctors and they are going to want to work with you,” he said.

The Vendor Selection Process

A significant amount of planning went into Duke’s process to select an imaging vendor partner. Part of that process were said it held “hash-out” sessions to figure out its strategy. Among the questions it considered were what it wanted to do with its imaging strategy, what it was looking for in a partner, and whether or not it wanted to manage its images or store them remotely with a third-party. (It decided to manage the images itself.)

In created a list of the 10 most important factors to consider in its request for proposals (RFPs):

  1. Completely industry standards based and devoid of proprietary standards.
  2. In-production integrations with cardiology and radiology PACS.
  3. In-production integration with the EHR.
  4. In-production capture of DICOM and non-DICOM into the archive.
  5. Spectrum of non-DICOM data stored.
  6. Data will be delivered back to the original endpoint in the method it was loaded.
  7. Anything DICOM stored in an archive can be accessed by any DICOM viewer.
  8. Independence from associated vended PACS.
  9. The ability to segregate data and restrict access to data as needed.
  10. Commitment to a quick build timeframe.

Roth said that the RFP process was worth it, but advised organizations to be skeptical of the vendor responses. He recommended doing additional research on claims of vendors. Duke eventually selected TereMedica.

Have an Inclusive Governance Structure in Place

A governance structure is critical to the imaging architecture, Roth said. “Any time you are going to have a fairly flexible architecture, you are going to want governance, and it’s really important to do this correctly,” he said. Governance will probably be led by large volume users, i.e., radiology and cardiology. Other members of the governance structure should be one or two “superusers” who are not from cardiology and radiology; smaller groups that do imaging occasionally; and representatives from video and telemedicine, IT, and compliance and risk management.

Roth cautions that the creation of an enterprise imaging strategy has the potential to devolve into imaging specialties against non-imaging specialties. He recommends checking politics at the door. Radiologists bring decades of experience and will be at the forefront of imaging because they do it every day, he said. “Radiologists may look at the transition as a threat because the stuff they used to have a monopoly on is going to enterprise,” he said. At the same time, non-radiologists, who are new to the table, come with interesting thoughts and workflows, but probably don’t posers a high level of knowledge. It’s important to be consensus builders, with the goal that “every image taken is going to have availability throughout your health system,” he said, adding that that all images have compliant metadata and access security.

Even with that groundwork laid down, there are still a lot of tough questions to be addressed Roth said: setting priorities; what processes, system interfaces and procedures; and who is accountable for maintenance and future. On the financial side, an organization needs to decide on either a pay-as-you-go model or image storage as the cost of doing business.

In summary, Roth said, “the archive we wanted is the one that would open up as many doors as possible while allowing us the flexibility to do what we wanted to do.”


Sponsored Recommendations

Healthcare Rankings Report

Adapting in Healthcare: Key Insights and Strategies from Leading Systems As healthcare marketers navigate changes in a volatile industry, they know one thing is certain: we've...

Healthcare Reputation Industry Trends

Navigating the Tipping Point: Strategies for Reputation Management in a Volatile Healthcare Environment As healthcare marketers navigate changes in a volatile industry, they can...

Clinical Evaluation: An AI Assistant for Primary Care

The AAFP's clinical evaluation offers a detailed analysis of how an innovative AI solution can help relieve physicians' administrative burden and aid them in improving health ...

From Chaos to Clarity: How AI Is Making Sense of Clinical Documentation

From Chaos to Clarity dives deep into how AI Is making sense of disorganized patient data and turning it into evidence-based diagnosis suggestions that physicians can trust, leading...