Over the past 10 years, the digitization of radiological studies has brought significant gains in both productivity and long-term cost savings through the elimination of paper and cellulose films in favor of compact disc and Web-based imaging.1,2 Simultaneously, the overall volume of diagnostic studies performed in American hospitals has nearly doubled over the same period.3 This has left neurosurgeons, who must make clinical decisions on these studies, with the unappealing task of managing extensive portfolios of digital imaging studies for a large fraction of patients in their practice, even those on whom they do not operate. Even though technical advances in radiology have transformed how neurosurgeons practice, these innovations present their own unique challenges in the modern medical setting.
Image management today is usually performed with a picture archiving and communication system (PACS). PACS is a collective term used to define a computer database that receives diagnostic studies from various imaging modalities (computed tomography, magnetic resonance imaging scanner, etc), archives those studies for future reference, and sends (communicates) those studies to viewing stations. This system serves as the backbone of all hospital radiology systems. Medical images stored on a PACS server are encoded and transmitted according to the digital imaging and communications in medicine (DICOM) standard.4
When viewing images archived locally on a hospital PACS server, the surgeon has access to the database through either a standalone application or a Web application designed and configured specifically for that server. However, when patients present to the ambulatory office, emergency room, or intensive care unit as an urgent transfer, their outside radiological studies now come in the form of a compact disc containing DICOM images. Although this is a convenient and inexpensive form of transmission, compact discs create several problems, each of which can cause delays that degrade a surgeon’s efficiency, interrupt clinical workflow, and ultimately result in increased costs to the medical system.
Compact discs, whether originating from radiology centers or other hospitals, come with a viewer application that allows the physician to view the accompanying imaging study. However, each vendor uses a different viewer, forcing the physician to constantly learn a new viewer interface. Some viewing applications lack essential tools (eg, cut lines) that make it difficult to properly interpret the study and/or use it for operative planning. Frequently, viewers that accompany a radiological study on compact disc require installation of software, which is often not permitted on general hospital workstations.5 Additionally, storage and sharing of compact discs is likewise tedious and time-consuming, often leading to a misplaced or lost compact disc.6 Finally, some compact discs or their accompanying viewing applications simply fail to load because of software incompatibility or physical deterioration of the recording medium.6,7
Sung et al6 investigated the rate of repeat imaging performed on emergency department transfer patients and quantified the reasons for ordering a repeat study. The authors found that between 14% and 25% of imaging studies were repeated because of information technology (IT) reasons, including inoperable compact discs, poor navigation tools, and difficult user interface. Additionally, survey respondents frequently cited the need for multiplanar reconstructions and the availability of a study within the hospital PACS as justification for repeating a study.
Current commercial PACS implementations lack the essential functionality of allowing quick, decentralized importation and archiving of these outside radiological studies. Therefore, in most if not all ambulatory offices, surgeons are forced to view, store, and share outside radiological studies in their original form on a compact disc. Besides the inherent problems of this practice as outlined above, the process of reloading a compact disc, storing it for future use (eg, surgery), and physically transporting it for any sort of collaboration significantly affects clinical workflow and degrades efficiency.
We sought to extend the benefits of a centralized PACS to the daily clinical operations of a busy academic neurosurgery practice. To address the problems of viewer heterogeneity, software incompatibility, workflow inefficiency, and the need for collaboration when viewing outside studies, we decided to implement a departmental image management system (PACS) to handle all outside radiological studies for both outpatients and inpatients. Because we had limited IT resources, our primary design goals were that this system would be easy to maintain and affordable and would provide secure, ubiquitous, and uninterrupted access to archived outside radiological studies. We also wanted to use readily available, open-source software that would be easy to use and navigate, would provide a consistent interface and tool set for all our studies, and would offer multiplanar reconstruction of cross-sectional images.
Hence, the primary goal of this study was to design and implement a departmental radiological image management system that would reduce inefficiency, improve clinical workflow, and allow for enhanced collaboration. The secondary aim of this study was to decrease overall costs by reducing the need for repeat imaging.
METHODS AND IMPLEMENTATION
The backbone of the system is composed of a Macintosh mini computer running Snow Leopard server 10.6 operating system (Apple Inc, Cupertino, California). We chose this system because it is easy to manage, inexpensive, and compact. The actual image database is stored separately on a Drobo S Redundant Array of Independent Discs (RAID 5 configuration) storage device that is connected to the server via a FireWire port (Drobo Inc, San Jose, California). The Drobo S device is redundant, meaning that the stored data are mirrored (duplicated) between hard drives. Hence, even in the event of a hard drive failure, the device ensures that there is no loss of data or interruption in service. It requires no maintenance and is easily expandable to allow increased storage capacity. Finally, the server is connected to the University of Pennsylvania internal network via Ethernet. Because the system stores sensitive patient information, the design process and implementation were performed in compliance with HIPPA and related institutional policies.
The open-source OsiriX 64-bit software allows this server to function as a PACS.8 Although this software has a limited tool set when used for the purpose of managing a large image database, its ease of use, reliability, and minimal maintenance requirements make it very attractive for this purpose. Online, Web-based viewing of studies residing in our database is facilitated by Weasis, a built-in OsiriX extension. This extension is automatically downloaded and executed on the client computer, without any need for configuration, when a user accesses the image database.
Finally, each ambulatory office is equipped with 2 loading stations. Each loading station is a Macintosh mini computer configured with OsiriX. In this case, OsiriX acts as a universal DICOM image viewer and extractor. The loading stations are also configured with Growl, an open-source notification system that alerts the user to the progress of image extraction and transmission to the server.
We have incorporated this system into our daily clinic workflow, which functions as follows (see the Figure). While the surgeon reviews the patient’s electronic medical record (EMR), the nurse practitioner, resident, or administrative assistant inserts the accompanying patient’s compact disc into the loading station. No other user intervention is required beyond this step. The system is configured to automatically extract the DICOM data from the compact disc and archive it on our PACS server. After this process is completed, a notification appears on the screen informing the user that the DICOM data has been extracted and archived. The compact disc is then automatically ejected and can be returned to the patient. This process is of course identical for patients who are admitted through the trauma bay or are transferred in directly to our intensive care unit. Because the simplicity and popularity of this system, we have installed loading stations at both of these locations also.
FIGURE. Clinical wor...Image Tools
To date, our system has been functioning for > 12 months and has maintained a 99% uptime. The initial deployment included 1 hospital and 3 loading stations; a second hospital was brought online 3 months later. Over this 12-month period, access to the database went from approximately 25 requests (online connections to read/write) per day to about 60 with no decrease in performance. Our image database currently holds > 17 000 studies totaling > 3 million images and uses 1.4 TB of hard drive space. Our failure to read and import DICOM data from an outside compact disc is approximately 2%, in which case the physical compact disc is retained according to previous practice.
The costs associated with implementation of this system are outlined in Table 1. The IT fee for initial setup was based on 40 hours of work at the cost of $125/h.9 In sum, the total cost of the system and deployment was $9870. Furthermore, we assume an annual maintenance cost of approximately $1974 (20% of the initial implementation cost10).
Over the same 12-month period that our PACS system was active, the Department of Neurosurgery at the Hospital of the University of Pennsylvania received 568 urgent transfers from hospitals outside our healthcare system. The Medicare fee schedule, which provides conservative pricing, sets the cost of computed tomographic and magnetic resonance imaging scans at $250 and $800, respectively. Therefore, breaking even on the initial hardware cost of this system and deployment would require preventing 40 computed tomographic scans, 13 magnetic resonance images, or a mixture thereof. According to Sung et al6 and Haley et al,11 approximately 4.5% to 7.5% of imaging studies are repeated for reasons that can be avoided with a robust image extraction and archiving system. At these rates, our implementation prevented between 25 and 42 repeat studies performed for IT reasons. The cost savings associated with avoiding these scans would, on its own, justify the initial cost of deployment well within a year.
We performed a similar analysis using the spine surgery database from our ambulatory offices. Once again, over the same 12-month period, there were 1913 unique office visits that resulted in surgical intervention. Approximately 80% of the patients seen in our ambulatory offices were referrals from another hospital system, of whom 80% had previous imaging performed and stored on compact disc (the remaining were on paper or cellulose or performed by a University of Pennsylvania affiliate). When the same repeat imaging rate is applied to this patient cohort, our system prevented between 54 and 90 repeat studies over 12 months, which translates to a cost savings between $43 200 and $72 000 (see Table 2).
We have found that a departmental implementation of PACS significantly improved efficiency and clinical workflow and enhanced collaboration in a variety of clinical environments. Furthermore, the system reduced overall costs by decreasing the rate of repeat imaging. In turn, this reduction of duplicate imaging led to decreased patient exposure to radiation, which continues to be a mounting concern among epidemiologists.12,13
Although the conversion to digital radiology is cost-effective in the long term and results in increased productivity, archiving of outside radiological studies by hospital radiology departments has so far proven to be an unfunded endeavor.1,2,6 At large referral centers, the storage, transport, and viewing of these studies on compact disc pose significant challenges to an efficient workflow. Except for a few examples in which radiology departments have streamlined the process of importing studies onto the institutional PACS and typically charge a fee for storage and formal interpretation of the study, the management of outside studies is a task most intuitions do not actively engage in.5 We have demonstrated that a simple and inexpensive computer system can provide a functionality that large commercial PACS do not have, namely an efficient, decentralized method for importing and archiving outside radiological studies. This system was easy to implement and over time has proven to be straightforward to maintain.
This implementation not only has allowed us to achieve cost savings but also has led to gains in productivity in our ambulatory offices, improved efficiency in the operating room when patients return for surgery, and fostered collaboration with surgeons from other departments. In our ambulatory offices, the surgeon no longer needs to wait for a compact disc to load or to learn how to navigate multiple DICOM viewing applications. With the Weasis extension, the user interface and tools for image analysis are always the same. Most important, with the latest iteration of OsiriX, we have been able to extract DICOM data from > 98% of outside radiological studies that arrive on compact disc.
Furthermore, the surgeon and the support staff are freed from the responsibility of maintaining and storing the compact disc itself. The surgeon can view the images on the loading station before seeing the patient or review the studies with the patient in the examination room through our online viewer. Once the study is loaded onto the server, collaborators (ie, other surgeons or radiologists) can view the images from their own computers. Before this implementation, simultaneous online review and collaboration was feasible only with studies performed in house and stored on the hospital PACS.
The same gains in efficiency have been realized for patients who go on to surgery. Instead of relying on physical compact discs, supplied by either the support staff or the patient, the surgeon can retrieve archived studies from the server. The surgeon does not have to keep track of the hard copies (compact disc) of the images; likewise, the patient does not have to bring the studies with him or her on the day of surgery. This not only provides a significant safeguard against the all-too-common lost or forgotten disc but also avoids the associated delays on the day of surgery, which can range from waiting for a disc to load to the acquisition of a repeat imaging study.
Additionally, OsiriX contains advanced, built-in tools to produce multiplanar reconstructions from DICOM data, even when they are not supplied on the compact disc as part of the original study. At our institution, we see and evaluate a large number of patients with skull base lesions for whom collaboration among otolaryngologists, radiologists, and neurosurgeons is necessary. This system allows us to quickly evaluate the outside radiological studies and to collaborate instantaneously with our neuroradiology and otolaryngology colleagues, obviating the need to arrange for physical transportation and sharing of compact discs. Furthermore, our system serves as a permanent repository that the surgeon can use to identify prior images for comparison when evaluating a new study for disease progression. Our neuroradiologists have also benefited from access to prior images and have welcomed this solution.
Finally, this system has also become an integral part of our evaluation and care of trauma patients. Our trauma service acts as a regional catchment center; patients arrive from a variety of referral centers and in various stages of management. These patients frequently arrive with radiological studies that must be reviewed and rereviewed by residents and staff physicians from multiple services. Our residents are able to archive these studies on our system so that chief residents and attendings can evaluate these studies in their offices or from home. Together with our colleagues in trauma, this solution allows us to quickly and collaboratively evaluate a patient and implement treatment with an efficiency and convenience that previously was not possible.
A common objection to this approach is that we have shifted the responsibility for management of these outside studies from the institution to our department, as well as the financial burden that comes along with that responsibility. This objection is entirely valid. However, current large, commercial PACS providers would not have addressed all of our needs. Although we have the ability to archive outside studies on the internal PACS at our institution, this process unfortunately takes 1 week after the appropriate requisition form is completed. This is not a problem unique to our institution and is frequently the result of administrative policies.6,7,14 Even if the turnaround time were reduced to hours, it would still be too long to improve the efficiency of the outpatient clinic and would be of no benefit to critically ill patients who are transferred to our intensive care unit and require an immediate assessment.
It is likewise true that from a broader IT infrastructure perspective, this implementation adds a level of complexity. Specifically, this system runs in parallel with the existing institutional PACS and EMR systems. It is important to note, however, that although on the surface this implementation duplicates a hospital’s PACS service, it also adds a critical functionality that was previously absent. As described in detail, this functionality has enabled our department to be more efficient when dealing with outside imaging studies. Our users have found that despite having yet another database to query, this solution is far superior to the previous practice of relying on compact discs. We have adapted the radiology section of our EMR to now include a study location field, which alerts the clinicians as to which database holds the relevant imaging studies.
Furthermore, integration of this system with hospital-managed systems is limited only by institutional IT policies, not technical feasibility. In fact, the OsiriX online viewing portal facilitates integration of imaging studies with other systems provided that the existing system allows customization. We continue to work with our IT department to integrate our search and imaging results directly with the hospital EMR and thus provide a more complete patient imaging portfolio. This will be especially important in 2014, when the stage 2 meaningful-use requirements of the Health Information Technology for Economic and Clinical Health Act go into effect. These requirements stipulate that to qualify for the Medicare and Medicaid EMR incentive program, > 10% of imaging results must be accessible through the EMR. With appropriate IT and programming support, integration of OsiriX into existing hospital systems is possible and straightforward. Our current efforts are focused on augmenting the DICOM extraction and archival process so that patient names and medical record numbers will be reconciled with the institution’s patient registry, therefore creating a complete imaging record that can be accessed from a central online portal.
The cost analysis presented has a number of limitations. We used a published rate of IT problems associated with imaging studies on compact discs and assumed that this rate was comparable at our institution. Although this assumption may be valid, we could not be certain of this unless we were to perform a similar analysis to that by Sung et al6 at our institution. This, however, was not the goal of the present study and would be difficult to ascertain now because opening imaging studies directly from compact discs has completely fallen out of favor in the department since the implementation of our system. Our analysis also assumes that each transfer patient had only 1 imaging study that needed to be repeated. Although this tends to be true for our spine patient cohort, the vast majority of our elective cranial cohort, as well as transfer and trauma patients, had > 1 imaging study performed at the referring institution. Hence, any IT issues related to a single patient would be multiplied, as would the cost, with each additional study. Our analysis also assumes that the cost of an imaging study is constant among providers and health insurers and is equal to the Medicare global fee schedule. This assumption will also underestimate the true additional costs of a repeat imaging study to the health system as a whole. Furthermore, because our analysis did not include the entire ambulatory office patient population who eventually proceeded to surgery, nor could we reliably factor in the frequency at which a compact disc is lost or forgotten, the potential cost savings presented are likely grossly understated. Finally, we are analyzing cost savings from a societal or payer perspective, not necessarily from an institutional or department perspective. Our department paid for this system and did not realize any direct cost savings through a reduction in repeat testing, although there were significant gains in efficiency and coordination of care (albeit harder to quantify), which we believe more than justify the cost. Moreover, although it is wasteful to repeat radiographic studies when a patient is transferred between institutions, in a fee-for-service model, the receiving institution may profit from these studies being repeated. Our analysis is not designed to account for these issues.
Finally, a number of commercial solutions on the market offer similar benefits (eg, lifeIMAGE, ClearCanvas Inc). One advantage of these solutions is their compatibility with the Windows operating system. Although a commercial solution could potentially be more robust and secure, as well as scalable, it would undoubtedly come at a significantly higher cost and longer lead time than our open-source, departmental implementation. The software cost alone for the ClearCanvas PACS is $15 000. Although this is a nominal amount for a large hospital system, the financial considerations of surgeons in private and group practice might differ, especially because these commercial systems do not offer any added functionality.
We do not propose that this implementation can replace or scale to the level needed to serve an entire hospital system. This was not the purpose of this study or a design goal during the development of this solution. The overarching goal of our implementation was to overcome the inherent inefficiencies when dealing with outside imaging studies in the outpatient and acute care setting by providing complementary functionality to the commercial, hospital PACS. However, we propose that this system could resolve inefficiencies on a daily, departmental level and, with proper support from the hospital IT department, integrate with hospital PACS/EMR systems to offer a more complete radiological patient record.
The evaluation and storage of outside radiological studies at large referral centers present unique challenges to a surgeon’s productivity. We have demonstrated that a departmental implementation of a PACS can significantly boost efficiency, improve workflow, and enhance collaboration. By relying on open-source components, this system was affordable and straightforward to implement and maintain. Finally, this system was effective in reducing costs by decreasing acquisition of repeat imaging studies for IT reasons.
The authors have no personal financial or institutional interest in any of the drugs, materials, or devices described in this article.
1. Bryan S, Weatherburn GC, Watkins JR, Buxton MJ. The benefits of hospital-wide picture archiving and communication systems: a survey of clinical users of radiology services. Br J Radiol. 1999;72(857):469–478.
2. Fang YC, Yang MC, Hsueh YS. Financial assessment of a picture archiving and communication system implemented all at once. J Digit Imaging. 2006;19(suppl 1):44–51.
3. Smith-Bindman R, Miglioretti DL, Larson EB. Rising use of diagnostic medical imaging in a large integrated health system. Health Aff (Millwood). 2008;27(6):1491–1502.
4. Bidgood WD Jr, Horii SC. Introduction to the ACR-NEMA DICOM standard. Radiographics. 1992;12(2):345–355.
5. van Ooijen PM, Guignot J, Mevel G, Oudkerk M. Incorporating out-patient data from CD-R into the local PACS using DICOM worklist features. J Digit Imaging. 2005;18(3):196–202.
6. Sung JC, Sodickson A, Ledbetter S. Outside CT imaging among emergency department transfer patients. J Am Coll Radiol. 2009;6(9):626–632.
7. Gupta R, Greer SE, Martin ED. Inefficiencies in a rural trauma system: the burden of repeat imaging in interfacility transfers. J Trauma. 2010;69(2):253–255.
8. Rosset A, Spadola L, Ratib O. OsiriX: an open-source software for navigating in multidimensional DICOM images. J Digit Imaging. 2004;17(3):205–216.
9. Gauger RH. Profile of IEEE Consultants. Washington, DC: IEEE USA. 2008;1–24.
10. Hillestad R, Bigelow J, Bower A, et al.. Can electronic medical record systems transform health care? Potential health benefits, savings, and costs. Health Aff (Millwood). 2005;24(5):1103–1117.
11. Haley T, Ghaemmaghami V, Loftus T, Gerkin RD, Sterrett R, Ferrara JJ. Trauma: the impact of repeat imaging. Am J Surg. 2009;198(6):858–862.
12. Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med. 2007;357(22):2277–2284.
13. Sodickson A, Baeyens PF, Andriole KP, et al.. Recurrent CT, cumulative radiation exposure, and associated radiation-induced cancer risks from CT of adults. Radiology. 2009;251(1):175–184.
14. van Ooijen PM, Broekema A. Including oncological image data from CD-ROMs into the normal workflow. J Oncol Pract. 2008;4(4):210–211.
The authors describe their development of a low-cost picture archiving and communication system (PACS) for the management of “outside images” made available on compact disc. Using consumer hardware and public-domain software (Osirix as local viewer and PACS server and Weasis as remote viewer), the department has created a robust and stable digital imaging and communications in medicine (DICOM) data extraction system and has combined it with a local and remote access and viewing system.
The system effectively extracts DICOM data from compact discs written with propriety software (98%), provides consistency in the user interface and tools, and thereby avoids the many problems associated with the variety of viewing tools and DICOM formats that manufactures include with their products.
The system handles, at the time of submission, 1.4 TB of data, the addition of 20 to 50 studies, and > 60 intermittent viewing connections daily. Of greatest importance, it provides all or some of a patient’s image data to the surgeon at the time of clinical assessment or operation.
The unavailability of the images on the institutional PACS was one of the reasons for the high rate of repeat imaging in the study by Sung et al.1 The system described in this work runs in parallel to the institution’s PACS. As a result, the longitudinal history of the imaging for the patient in either system is likely to be incomplete. Ideally, the institutional PACS should be configured to accept imported image data from this system so that a complete longitudinal data set is available for patient care.
The system developed also provides a repository for teaching studies.
Although some health systems use image brokering systems to interface between digital imaging devices (digital radiography, computed tomography, magnetic resonance, ultrasound, digital angiography) and institutional PAC systems, there will always the need for “portable data transfer.” The system described here is a cost-effective, robust, uncomplicated, reliable strategy if importing DICOM data into a PAC system is cumbersome or not possible.
David Douglas Cochrane
1. Sung JC, Sodickson A, Ledbetter S. Outside CT imaging among emergency department transfer patients. J Am Coll Radiol. 2009;6(9):626–632. PubMed Cited Here... |
Impactful use of healthcare information technology requires the deployment of such tools within effective and efficient clinical workflows. Although a departmental picture archiving and communication system (PACS) system as described in this article offers great value in preventing unnecessary imaging, enhancing access to critical patient data, and delivering of cost savings, there is concern about how parallel systems---departmental PACS and institutional electronic medical records---affect clinical workflow and their downstream effects on patient safety and quality of care. In deciding to deploy a departmental PACS, careful attention must be paid to designing an effective clinical workflow that delivers expected results. Allowing the clinical workflow to evolve on its own presents great clinical risk to patients and financial risk to institutions.
Barry P. Chaiken