No Image Left Behind

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
Screenshots courtesy of National Center for Biotechnology Information, Montefiore Medical Center
In the good old days, image management was an insiders' club largely comprised of radiologists and PACS administrators well-versed in the nuances of DICOM, modality worklist and imaging informatics. Today, however, image management is hurdling toward an enterprise model characterized by broader, looser boundaries and lots of data.

What's driving the new model? "At the end of the day, someone in Washington is going to say a picture is just as meaningful as data, and [an enterprise image record] is meaningful use," predicts Lou Lannum, director of enterprise imaging at Cleveland Clinic. (And someone in Washington has. "We can't have an effective electronic health information system that can't move images," David Blumenthal, MD, former national coordinator of the Office of the National Coordinator for Health Information Technology said in January.)

A centralized enterprise archive provides a platform for regional health information organization (RHIO)- or health information exchange (HIE)-based imaging sharing, says Mony Weschler, director of clinical ancillary systems and emerging health information technology at Montefiore Medical Center in New York City. Departmental archives make it difficult to replicate and share data.

Economics and technical developments also are fueling the enterprise approach. Multiple departmental archives are costly to deploy and manage. Finally, new software applications create DICOM wrappers for previous imaging outliers like endoscopy and ophthalmology.

Toward the enterprise image record

Cleveland Clinic epitomizes the transition to an enterprise model. Early in 2011, the health system promoted Lannum from administrator for radiology imaging informatics to administrator for enterprise imaging and tasked him with locating images across the enterprise. The ultimate goal? Build a longitudinal imaging record to correlate with the EMR.

Lannum and his team are taking an asset management approach and inventorying imaging including dermatology, pathology, endoscopy and point-of-service imaging.

The team opted to focus on easy images first. These "x-ologies," in Cleveland Clinic parlance, are equipped with an intermediate device that can produce an outbound DICOM message. Such systems can be interfaced with an EMR, and patient demographics are automatically shared and reconciled. Initial targets include women's health, ophthalmology and the Digestive Disease Institute, which produces endoscopy images, and C-arm datasets.

Montefiore Medical Center has solved earlier challenges of integrating its endoscopy imaging data into PACS; the center has transferred about 40 to 50 percent of the data into PACS. "Most PACS don't yet natively accept endoscopy data," shares Weschler, "but there are solutions that create a DICOM wrapper around the video data. It's no longer the challenge it was two years ago." Some new endoscopy cameras output DICOM natively.  Alternately, a few third-party systems provide the middleware between modalities and PACS, and some PACS provide the DICOM wrapper.

As technical issues are conquered, human issues take center stage. It's critical that providers in image-generating departments adapt workflow to image management and generate an image order to link the patient to the image. If an image is put into the archive without normalizing the demographics with the EMR, it's nearly impossible to track the image. "This isn't just about storage," Lannum says, "it's about storage and subsequent access." Image retrieval requires that workflow is modified in a way that allows matching between imaging metadata and the EMR, which ultimately enables future viewing via the archive.

Departments outside radiology, however, don't necessarily generate image orders. A dermatologist might decide to take a digital photo at the end of the appointment and request storage. Without an order or accession number, there is no link for retrieval.

Montefiore employs a closed loop PACS to address the retrieval issue. All orders go through computerized physician order entry (CPOE). They are generated in the EMR and sent to the RIS, which populates the PACS and provides the modality worklist. As radiology, cardiology and endoscopy images are acquired, they are profiled and matched to the orders to guarantee patient integrity, interpreted on the PACS and resulted via voice recognition or exception-based reporting in the RIS or CVIS. Finally, the reports are sent back to the acute care EMR, ambulatory EMR and Clinical Looking Glass, an internally-developed decision support program for outcomes analysis.

The Hospital of the University of Pennsylvania in Philadelphia has developed a web-based solution that circumvents DICOM and workflow challenges. Digital or digitized images such as endoscopy, ophthalmology and echocardiography studies are keyed into the system and can be retrieved via patient name or medical record number, explains Steven C. Horii, MD, director of medical informatics in the department of radiology. Images and reports are linked by the date of the exam.

Storage drain ahead: Pathology

Pathology rests on the edge of the enterprise archive precipice. Cleveland Clinic is taking a two-pronged approach starting with digital photographs of gross anatomy. A DICOM wrapper will be added and images will be archived, explains Lannum. The health system will likely use a similar workflow for dermatology images.

The bread and butter of pathology—whole-slide imaging—represents an entirely different story. The challenges start with massive storage needs. Weschler confides, "Pathology can generate as much data in one year as the rest of system in seven years." He estimates pathology's annual data output at 125 to 150 terabytes (TB). High density storage and dropping prices partially address this challenge.

The more complicated half of the challenge is digital slide acquisition, says Weschler. The center is working with a vendor to develop a scanner to digitize slides, a process that will likely take at least two years. Workflow and economics also are part of the discussion. Unlike in radiology, where digital acquisition replaces film, digital pathology requires a physical slide. Nevertheless,  the transition could boost savings by improving pathologist workflow and eliminating the need for surgical pathologists to travel among campuses.

The radiology and pathology departments at the Hospital of the University of Pennsylvania, which has a slide scanner and software that links the radiology and pathology databases, have started to discuss limited archiving of pathology datasets, with interesting and unusual cases archived in the radiology PACS. Horii does not envision a wholesale transition to PACS for the pathology department. "The workflows are different enough and use of images is different enough that it's going to be very difficult to build a PACS that does everything," he explains.

The burgeoning list

Montefiore is tackling the enterprise image management issue from multiple angles. Take for example dentistry. The center has digitized dental images across its four practice sites and converted images from a non-DICOM JPEG format to DICOM. Images are stored in central servers and are ready to be consolidated into the enterprise PACS as time and budget allow.

Another image-generating department, ophthalmology, has proven slightly more challenging. While fundus images are fairly straightforward, ophthalmologists also rely on multiple instruments that produce hybrid data—images and tabular data or numbers. "It's not unlike echocardiography," explains Weschler, "we have to combine images and data and present them to the users on a workstation with images and data pre-populated in the report." The industry is moving toward DICOM standards; however, non-DICOM legacy equipment is the norm, continues Weschler.

The Montefiore plan illustrates the complexity of external 'ologies. Ophthalmology images will be stored in the enterprise PACS, where referring physicians will view images. The center also plans on middleware: an ophthalmology workstation to display images and data for ophthalmologists.

Despite the complexities, Weschler recommends an enterprise approach to PACS. Image management fundamentals overlap the 'ologies, and an enterprise team can identify the overlaps, leverage infrastructure and build a robust, cost-effective enterprise archive. Horii, however, emphasizes the utility of the web-based application. "It's a tradeoff. The enterprise PACS can deliver better performance, but it sacrifices ease of use for non-radiology departments and entails a larger client."