 Gary J. Wendt, MD, MBA |
Picture archiving and communications systems (PACS) typically are deployed within a radiology department to help improve workflow and decrease cost, as well as enable these images to be distributed throughout the enterprise. PACS combined with voice recognition allows very rapid turnaround of radiology examinations and is a proven benefit. The technology has matured and is now relatively commonplace within the radiology department and for enterprise-wide distribution and linking into the electronic medical record (EMR). While this is proven technology, the PACS has much more potential and can be extended into other imaging areas across the enterprise. There are many other areas of image acquisition across an enterprise, and currently most of these are not captured, or if they are captured, they often are put into a proprietary system resulting in multiple isolated silos of information. This is inappropriate as it increases the cost of providing care and does not lead to universal enterprise-wide image access. Incorporating these images into an existing PACS solves both of these problems: a more appropriate term for a system such as this would be enterprise imaging management system. This allows the system to be dissociated from radiology and any potential turf tissues that may exist.
A HISTORY
The University of Wisconsin (UW) implemented its initial PACS in 1997, with the priority of improving patient care by reducing lost films, improving report turnaround time, and allowing patients to immediately return to the clinic after imaging examination due to the ability to have images open at multiple locations at one time. This was successful, and the focus of the PACS expanded to include linkage into the EMR as well as more universal enterprise-wide distribution of images. In 2006, the UW system is virtually 100% filmless with the exception of some mammography and small outlying clinics. This progression is a relatively standard evolution of the imaging chain progressing from single modality application, to radiology-wide application, and then progressing to enterprise distribution, as well as linking into the electronic medical record and HIS/RIS to provide an integrated imaging workflow solution. When evaluating the overall workflow solution provided by PACS at the University of Wisconsin, one of the areas of focus was the reduction in film use and potential cost savings from this reduction. During the analysis, it became apparent that while film reduction does remove cost, the major expense that is eliminated (reduced) is the cost of moving and storage of objects (film jackets).
The next portion of the PACS expansion at the UW sought to capture other images that were used and acquired throughout the enterprise. Some examples of these images include endoscopic images generated in gastroenterology, pulmonary, and ENT clinics; pathology images; and photographic images generated in dermatology and ophthalmology. Several clinics were identified for initial trials because they had equipment that already had an appropriate video output signal. GI and pulmonary medicine were used for an initial proof-of-concept trial, and subsequently the ENT clinic was chosen for full point of care DICOM image capture in all examination rooms. Pathology is also in the process of implementing direct DICOM image capture. During the initial trials, it became apparent that many of these clinics could benefit from the same object movement cost reduction by reducing or eliminating the “shadow charts” that often exist to handle these nonstandard images.
STEPS TO EXPANSION
 Figure 1. Point of care DICOM image capture bronchoscopy touch screen in use at the University of Wisconsin provides the ability to specify the precise location of any abnormalities. (Click the image for a larger version.) |
There are several steps that must be taken to effectively expand PACS into an enterprise image management system. The first is that there must be an understanding that this management system is a more effective tool for distributing images throughout the enterprise than the electronic medical record. An image management system understands the concept of like priors, so that in radiology, for example, a head CT is compared to a head CT. Similarly, outside of radiology, in dermatology, for example, a mole on the right forearm would be compared with a similar image from a prior examination. Simply placing images into an electronic medical record does not allow this flexibility and side-by-side comparison with priors. The next step is understanding the value proposition of the enterprise medical image archive: the PACS is already in place and is a significant investment with a proven return. Expanding this vision of radiology PACS into other clinical areas as well will allow institutions to realize the proven benefit to radiology PACS as well as improving the effectiveness of clinical care teams by providing a single point of image display and comparison. This leads to improved patient care by providing a more integrated medical record as well as improved speed of care delivery by allowing easy access to all images at any location in the enterprise by any care provider (or providers). Likewise, there is a single system infrastructure that provides a lower cost of ownership and maintenance: a single management and support team, allowing a single user interface and training session for clinicians, as opposed to multiple stand-alone image silos that lose all of these benefits.
CREATING A ROADMAP
After the value proposition is defined, a roadmap needs to be created for extension into the enterprise. There are many areas to look at in each enterprise, and each enterprise is unique. One needs to understand where technology is in all areas of the enterprise and to make sure that equipment is capable of digital image capture as it is purchased. The roadmap should also have a strong push to distributed electronic integrated images and reports that are linked into the EMR, along with the concept of capturing and storing images in a standard fashion using the framework provided by DICOM and Integrating the Healthcare Enterprise (IHE) as much as possible. Image islands and proprietary systems that would increase the cost of providing service should be avoided. The ideal vision would provide multiple unique working environments that are tailored to meet the specific needs of each area: for example, radiology, cardiology, and other visible light imaging areas. All should be linked into a common institutional image storage archive with subsequent links into the electronic medical record and bi-directional links into the hospital information system, radiology information system, and other scheduling information systems. These links are particularly important, as it is essential to provide modality worklist to ensure that all demographics are valid and provide correct linkage from the point of acquisition of the examination, through the system and out to other portals in the enterprise.
 Figure 2. Multimodality visible light images comparison screen provides the radiologist with a more complete patient history with both sinus CT and prior endoscopy images, so that a nasal mass that was not commented on in the endoscopy report can be ruled out. (Click the image for a larger version.) |
Another one of the challenges of a full enterprise imaging archive is addressing all areas in which film or other images are used, specifically the operating room, physician offices, and conference rooms. In addressing these areas, it became apparent that simply capturing visible light images using standard methods such as a scanner or additional camera and subsequently importing them into the PACS were inadequate. Using this approach led to several problems: At the time of capture, the file names that are generated often have no relevance to the actual patient, body part, or specific location being examined. If there is a significant delay between the time of image capture and the time it is imported into a DICOM format, the possibility of misidentifying the image increases dramatically. This problem is compounded if multiple patients are combined on a single piece of transfer media, opening up the possibility that images are placed into the wrong patient’s record. In addition, if the person capturing information is not the person importing and appropriately annotating each image, the possibility of misidentification is very high. In addition, if images are imported without appropriate location information (ie, precise location in/on the body), they are of little use to referring physicians and also will be of little use in the future for comparison to new examinations as it is impossible to correctly match images between new and old examinations.
Doing a direct DICOM point-of-care image capture provides the most effective solution for visible light imaging. A sample image of the touch-screen input device for a bronchoscopy examination (Figure 1) illustrates that these screens are customized based on examination and body part being examined. They allow the physician to quickly identify the exact site of image capture during the examination. Additional information, such as more specific locations or tumor grading information, can also be included as desired. A common argument against using dedicated capture boxes is that they are significantly more expensive than a digital camera or scanner. While this is true, one must consider the cost of additional physician/staff time and potential for error using the standard digital camera or scanner approach. With direct point-of-care DICOM image capture, there is no need to spend additional time to identify, import, and subsequently annotate DICOM images and send to the archive. In a clinic environment where a patient encounter may last only 15 to 20 minutes, adding an additional 5 to 10 minutes of physician time results in a significant decrease in throughput, additional physician workload, as well as a significantly increased probability of errors.
ALL FOR ONE
 Figure 3. This pathology image perfectly underscores the importance of scaling the PACS archive to accommodate this information-intense category: at 34K x 28K, the image is 2.9 gigabytes. (Click the image for a larger version.) |
Integrating all medical images into a common enterprise image management system provides multiple benefits. One is improved patient satisfaction as it allows the physician and patient to interact and review results of the examination in an office setting, a much more comfortable environment than trying to provide an interactive dialogue during an uncomfortable endoscopic procedure. This system also provides uniform access to all images for multidisciplinary care conferences so that multiple subspecialists can view all images associated with the patient. For example, radiology, endoscopic, and pathology images can be correlated during a tumor board conference. The system also allows improved patient care and reduction of follow-up costs by allowing immediate comparison during an examination. Figure 2 provides an example of a visible light multimodality screen and how it impacts patient care. Without access to the endoscopic images, a radiologist would normally need to refer to the endoscopic report in the EMR if available. If the nasal mass was not commented on, then a follow-up call would need to be made to the referring physician and the patient would possibly need to be notified that either a follow-up endoscopic imaging procedure or follow-up x-ray procedure would have to be obtained to rule out the possibility of a malignancy. With the current system at the UW providing direct point-of-care image capture, it is very simple for the ENT physician to obtain images during an endoscopic procedure, even of a normal-appearing polyp that was annotated and available for the radiologist to see during the subsequent sinus CT. This allowed a report to be immediately generated on the sinus CT that the examination was unremarkable with the exception of an incidentally noted polyp that was also seen on a prior endoscopic procedure. This eliminates a significant amount of time that would have been spent on both the radiologist’s and referring physician’s part to provide a follow-up phone call and possibly subsequent follow-up examination. It also eliminates any unnecessary anxiety generated by having to call a patient back for the evaluation of a potential malignancy.
PROCEED WITH CAUTION
 Figure 4. A pathology image is viewed by a user who has accessed the University of Wisconsin PACS (McKesson) from the Web. (Click the image for a larger version.) |
Although there are many compelling reasons to implement direct DICOM point-of-care image capture and a single enterprise image archive, there are several caveats that must be mentioned. The first is that the institution needs to evaluate all potential sources of image acquisition and do a realistic volume estimation, realizing that image volumes outside radiology can be significantly more than that currently generated within the radiology department alone (Figure 3). This necessitates scaling server and storage capacities appropriately. In addition, there are unique user interface requirements and image capture screens that must be appropriately defined and agreed upon by users at the site. One of the most significant hurdles is that of generating a modality worklist so that the device has the correct demographic information as well as appropriate examination-type information to eliminate manual information entry. This can often be accomplished by using current scheduling systems if they are capable of generating an HL7 feed to provide data to the enterprise archive if it is capable of generating a modality worklist or providing this information to a modality worklist device. The point-of-care image capture workstation can subsequently query this modality worklist to generate correct demographics. The step is also critical to allow effective electronic medical record integration, as there needs to be a unique identifier assigned to each examination, commonly call an accession number.
There also are some limits in the current standards as DICOM and IHE are in the early stages of establishing standards and frameworks outside radiology. These efforts include: the DICOM working group 26 in pathology; IHE Japan in implementing the visible light; and IHE France and Japan on developing the pathology profiles. Adherence to the standards, however, is critical as they will provide a method of open communications across multiple device vendors and institutions, so it is essential to insist on DICOM-compliant new devices as they are acquired.
Enterprise image management systems, in order to be effective, must also evolve from a picture-centric architecture to an object-centric architecture that will incorporate multiple items, including images, audio, scanned documents, and CAD data, for example. Deploying an effective enterprise image management system will allow an institution to realize many rewards. One of the most important benefits of direct DICOM capture at the point of care is the ability to effectively compare like prior examinations and images to a current examination. So, for example, a melanoma follow-up that has multiple lesions can be quickly compared rather than having to manually match each image each time a follow-up examination is obtained.
Other benefits include more effective subspecialty utilization. Departments will be able to create a larger virtual department in an area where there were previously multiple smaller practice environments. This will allow physicians more freedom to practice in the environment they desire: subspecialists can narrow their focus by receiving examinations from a larger patient base, while at the same time, generalists can continue to practice in clinics in community hospital environments while having easy access to the subspecialty consults. Another high-impact area is that of oncology. This is a field that requires input from many areas, including radiology, endoscopy, and pathology, as well as the consultation of multiple subspecialists to provide a team care approach to patient therapy. This provides a high value to clinicians and hospitals in offering an improved time to treatment, while at the same time improving value to patients by providing a best practice approach to care.
In summary, expanding PACS to an enterprise image management system is very cost-effective and provides improved patient care. To provide optimal benefits, however, it requires direct point-of-care DICOM image capture as well as full DICOM modality worklist, and neither of these is an insignificant hurdle to overcome.
Gary J. Wendt, MD, MBA, is a neuroradiologist and assistant professor of radiology at the University of Wisconsin-Madison.
