Dr. Dwyer is Professor of Radiology at the University of Virginia Health Sciences Center in Charlottesville, VA. He is also a member of the editorial advisory board of this journal.

T he concept of picture archiving and communication systems (PACS) was introduced during the period of 1978 to 1980. PACS is an electronic system for an image management infrastructure. Use of PACS in the radiology department is related to many functions, namely acquisition of medical image data; transmission of medical data and images via a network; display and manipulation of medical images; archiving of image data; interconnection of hospital information system (HIS) and radiology information system (RIS) data; and maintenance of security requirements to adequately protect medical data and images from unqualified seekers.

When seeking to implement a PACS into your department, a series of issues must be considered. The following sections will provide a basic overview of the relevant questions to ask in preparation of system acquisition and implementation.

Question 1: Do we really need a PACS?

Obviously, the first step in evaluating the need for a PACS is to determine if such a system can satisfy the necessary requirements in your department or institution, and to identify areas where the system can have the most impact on workflow and data management. To do this, we must identify those areas in which PACS has made the most impact.

A PACS increases department and hospital productivity by electronically managing digital image data. This is accomplished by the provision of an efficient electronic image distribution system which handles both intra- (within the hospital or complex) and inter-facility (between hospitals and remote clinics) requirements. Worklist management engines, consisting of a computer connected to the PACS network, carry out the programmed, preset rules for distributing image data among workstations according to such parameters as physician specialty, workstation location, staff events (i.e., coverage for absent staff members), and equipment down-time.

PACS provides an efficient means for archiving and retrieving image data. The increasing amount of digital modality image data being generated (such as 220 CT scans or 400 MRI scans per patient) requires efficient archiving, reasonably priced archiving media (such as digital linear tape), and efficient retrieval functions from large image databases. For example, a radiology department providing service to a 500-bed hospital will generate over a terabyte (1 trillion bytes) of image data per year. While the requirements for a scalable archiving system (at storage of 1 terabyte per year), in addition to those of a structured query language (SQL) for enhancing the retrieval functions, are difficult to satisfy, today's PACS technology is well equipped to handle such complex tasks.

Viewing of image and medical data generally is accomplished with soft-copy displays (interactive gray-scale monitors). Such monitors are the electronic display screens of PACS stations. Filmless implementation implies that the radiologist will be satisfied reading examinations from soft-copy displays, which often can be difficult to use, as they are designed by engineers and not those who do the actual reading; however, recently these workstations have begun to be designed by radiologists, resulting in improvements of the user interface.

One of the most significant issues in providing acceptable display protocols is the use of graphical user interfaces (GUIs). GUIs provide tool kits by which to modify the way images and data are displayed. Navigational tools implemented on soft-copy displays have gained user acceptance. Examples of such navigational tools include a stack mode display protocol, such as that used for CT and MRI displays. Stack mode display protocols involve the electronic stacking of CT images (each image is 512 * 512 * 12 bits), in which the user views the sectional images a slice-at-a-time and an interactive device enables the user to move through the stack at adjustable rates. In the case of MRI (256 * 256 * 12 bits), the stacking display monitors link display stacks (T1-, T2-, and proton density-weighted images). Such navigational tools were first introduced by Paul Chang, MD, University of Pittsburgh. Similar navigational tools are now being developed for digital mammography images (4.8 K * 6.4 K * 16 bits). The current size digital matrix for a display monitor is 2.5 K * 3 K * 8 bits.

PACS are highly efficient. Dr. Eliot L. Siegel, Director of the Department of Radiology and Nuclear Medicine, Baltimore Veterans Administration Hospital, MD, conducted a study of that institution's PACS and found the following significant benefits: Radiologist productivity increased 71%; lost or misplaced examinations declined from 8% to 1%; repeat studies declined from 5% to 1%; time required for a technologist to complete an examination declined by 60%; and report turnaround time declined dramatically.

In all, PACS has demonstrated three major advantages over traditional hard-copy reading and storage. First, an efficient medical image and data distribution system is provided. Images and data can be transmitted to physicians, both intra- and inter-site. Second, electronic archiving and retrieval of image data is made possible. Third, the image and data distribution system provides a mechanism by which interactive consultations may be accomplished between radiologists and other physicians.

Question 2: What technology do I need prior to installation and use of PACS?

The technology needed for a PACS is tied to the functions of a PACS and the applications to be accomplished.

There are six basic functions of a PACS, implemented by the rapidly changing technology. First, the function of digital image acquisition requires interfacing the PACS to the selected digital imaging modalities (CT, MRI, CR, DR, DF, US, film digitizer). The modality interfaces require that the devices to be used comply with the digital imaging and communication in medicine (DICOM) standard. As defined by Steve Horii, MD, University of Pennsylvania, DICOM is an engineering information blueprint defining modules of information and control instructions to exchange these modules. DICOM is a changing standard, with an expanding domain of syntax and information definitions.

Second, the network function of PACS requires both a local area network (LAN) and a wide area network (WAN). The LAN is used to transmit between nodes on the network for both image and data; it satisfies the intra-connectivity requirements. A WAN is used for inter-connectivity, such as to homes, clinics, distant hospitals, and academic medical centers.

For intra-connectivity, unshielded twisted wire pairs are very reasonable in cost of installation and maintenance. Data rates on switched LANs are approaching 1 gigabit per second. WAN's data rates range from 128 K bits per second (ISDN) to T1 (1.5 M bits per second) to the asynchronous transfer mode (ATM) of 155 M bits per second. The technology for LANs is rapidly changing due to the implementation of technology such as Web client-servers and digital subscribers' lines (DSLs). The network functions of any PACS, both LANs and WANs, changes rapidly with the constant improvements being made, resulting in lower hardware costs but increased costs in keeping up with the latest software (developed every 8 to 12 months).

Third, interactive gray-scale workstations, the so-called soft-copy displays, are nodes connected throughout the PACS. Image fidelity of such a display is measured by the physical characteristics of luminance levels, dynamic range, distortion, resolution, and noise. Psychophysical measurements, the result of receiver-operator curves and phantom contrast-detail patterns, affect image fidelity as well. Standard gray-scale workstation functions are: contrast manipulation, zoom and pan manipulation, scrolling, image orientation, magnification, gray-scale inversion, pertinent data display, a patient selection list, and a patient selection mechanism. Inclusion of these navigational tools is critical to the speed in which images can be read by radiologists.

Fourth, the selection of archiving technology depends upon archiving media and architectures. Archiving media may be in the form of magnetic discs, a redundant array of inexpensive discs (RAID), optical discs, or magnetic tape. Magnetic tape (DLT) is the favored medium due to its inexpensive costs and high reliability. Archiving architectures could be local, centralized, distributed, or Web-based. The direction of archiving architectures is towards centralized archiving, DICOM and HL7 interfaces, DLT tape media, and SQL databases. Other issues in archiving technology include backup, security, data mining, and knowledge discovery.

Fifth, the HIS/RIS data interface into PACS is critical to the success of the electronic image management infrastructure. HL7 (Health Level 7) was founded in 1987 to develop standards for the electronic interchange of clinical, financial, and administrative information among independent healthcare computer systems. HL7 was developed for hospital, information, clinical laboratory, enterprise, and pharmacy systems. This is the standard for electronic data exchange of medical data sets, and it is supported by most large hospitals.

Sixth, security requirements for a PACS are just now evolving. The Department of Health and Human Services has proposed standards for the security of individual health information and electronic use by health plans, healthcare clearinghouses, and healthcare providers (Department of Human Services, 45 CFR, Part 142, HCFA-0049-P). Regarding this security standard, the question is whether or not medical images are part of the patient's medical record. This security standard, proposed to become law in the latter part of 1999, consists of five sections:

1) administrative procedures to guard data integrity, confidentiality, and availability (e.g., a formal mechanism for processing records and for information access control including access authorization, establishment, and modification);

2) physical safeguards to protect data integrity, confidentiality, and availability (e.g., assigned security responsibility, media control, physical access controls, policy guidelines on workstation use and location, and security awareness training);

3) technical security services to guard data integrity, confidentiality, and availability (e.g., access control, data authentication, and entry authentication);

4) technical security mechanisms to guard against unauthorized access to data that is transmitted over a communications network (e.g., communication/network control); and

5) electronic signatures (for user authentication, non-repudiation, message integrity, and independent verification).

The amount of equipment required in a PACS installation depends upon the intended applications. It is best to implement architecture in well-defined, smaller sizes such as an intensive care unit (ICU)/Emergency Department (ED) image PACS, or a PACS for individual modalities such as CR, DR, laser film digitizers, ICU soft-copy display workstations in the Emergency Department, gray-scale workstations in the ICU, gray-scale workstations in the radiology department, an archiving unit with a database, and a workflow engine. These units would be interfaced to a LAN network.

Question 3: How do I maintain the PACS equipment?

PACS are computer systems. As such, they are made up of computer hardware, software, and networks. Maintenance charges vary from vendor to vendor. One vendor may charge a license fee for each user, while another vendor may charge a flat yearly price for the software of the PACS. A major difficulty in maintenance of a PACS is that the hardware and software will likely need to be updated every 8 to 12 months.

In deciding whether to implement PACS, an adjustment must be made to the idea that the initial cost of a PACS is just that--an initial cost. Subsequent charges will occur every few years in the continued operation and upgrading of PACS. Another difficult charge of operating a PACS is the pressure of scalability, in which others outside your original parameters will likely want to be included in the use of an initial PACS.

An additional consideration is that PACS will require staff support. An individual is needed to serve as the PACS manager, and additional assistants will be required to maintain such facets of PACS operation as quality control of the gray-scale monitors (expected life of display monitors is 2 years). The continual increase in acquiring digital imaging data means that the archiving media each year will increase in size, thereby increasing staff requirements.

Question 4: Are PACS vendors stable?

Unfortunately, the history of PACS vendors has reflected a lot of start-up companies and many mergers. This partly reflects the rapid change in technology, such as Windows NT systems and Web distribution systems. There will always be some level of uncertainty in acquiring and operat-ing a PACS. While it is a reality that PACS vendors are undergoing changes in both ownership and technology, being aware of the problem offers the best possibility of dealing with vendor and technology changes. Several pit-bull clauses (clauses which state that if the vendor company is sold, the new vendor must honor the original agreement) in purchase (or lease) of a PACS could go a long way for protection from uncertainties, though it is tough to get a vendor to agree to these. It is not clear at this time if a PACS consultant will be beneficial in dealing with uncertainties in acquiring a PACS. AR