The PC-based software package FEASIBLE (Feature Extraction And System Inspection By Logged Experiment) is an aid in designing, running and analyzing psychophysical experiments to evaluate the diagnostic accuracy of PACS components. The package has been developed within the scope of the IMAGIS project and is distributed non-commercially by the BAZIS organization. Until now the program has been used in three different kinds of studies: -The evaluation of digitized and monitor displayed images versus original films. -The evaluation of newly developed X-ray acquisition techniques versus the traditional equipment. -The evaluation of data-compression algorithms. Until now 8 different studies have been performed with the aid of FEASIBLE. For a few of these studies we will present the experimental procedure, some of the results, and the way FEASIBLE was used. Future extensions to the package will be discussed.

For PACS to be clinically acceptable and cost-effective, it must be able to perform the electronic equivalent of all the functions currently performed by the filerooms of conventional radiology departments. This performance must be at an equal or superior level to that of the conventional fileroom if PACS is to be welcomed in the clinical environment. It follows, then, that it is important to analyze each of those functions in detail. This paper set out to examine one of those, the retrieval of the appropriate prior studies needed for comparison with a new imaging study with the goal of answering two questions: (1)what are the quantitative implications of this function for image traffic and archiving in PACS? (2)do simple algorithms exist that would, in the large majority of cases, duplicate (or improve on) the actions of a conventional fileroom?

The objective of the studies was to evaluate the diagnostic performance of the image digitizing distribution and display equipment of the installed PACS (Philips) in the Utrecht University Hospital (see also procee-dings SPIE 914, 1988). The diagnostic accuracy was evaluated by means of observer performance studies, using ROC analysis. Three series of films were used: chest phantom radiographs, clinical mammographic films, and clinical head CT's. We compared: original films versus images digitized with a video scanner, and/or original films versus images digitized with a laser scanner. Both scanners have 1024x1024x8 bits output resolution. Four radiologists scored the presence on a 5-point scale. The results of the mammographic and head CT series will be presented in this paper. The results of the diagnostic accuracy studies of the chest radiographs and their specific methodology will be presented in a companion SPIE paper. Goal of the clinical efficacy study is to see if the use of a PACS multimodality viewing station at an internal ward has clinical advantages as compared to the conventional situation. Using PACS integrated with a HIS (including a RIS) more easier and faster information is accessible to the clinicians. In this paper we will describe: the procedures which are needed to evaluate the use of the workstation installed at the internal ward, the development of the user interface with four layers of the Image Management System, and the methodology of evaluation the clinical advantages.

The PACS project at Kyoto University Hospital in 1988 is reported in this paper. The old PAC system (KIDS: Kyoto univ. hosp. Image Database and communication System) developed since 1985 was reorganized and has been expanded to include several major digital imaging modalities such as X-ray CT, MRI, DSA and CR. The purpose of the new system (new KIDS) is to achieve the film-less environment in the field of neuro-radiology and evaluate the feasibility of the hospital wide PAC system. System configuration and present status are described. The newly designed fiber optic local area network and the new high-speed image workstation with 6 CRT screens are also mentioned.

The central concern of radiologists when evaluating Picture Archiving Communication System (PACS) is the diagnostic performance of digital images compared to the original analog versions of the same images. Considerable work has been done comparing the ROC curves of various types of digital systems to the corresponding analog systems for the detection of specific phantoms or diseases. Although the studies may notify the radiologists that for a specific lesion a digital system may perform as well as the analog system, it tells the radiologists very little about the impact on diagnostic performance of a digital system in the general practice of radiology. We describe in this paper an alternative method for evaluating the diagnostic performance of a digital system and a preliminary experiment we conducted to test the methodology.

A picture archiving and communication system (PACS) module for CT, MR, and US images is being developed in our department. The PACS module is composed of three subsystems: (1) image acquisition, (2) data management and (3) image display. A digital communication network links these subsystems into a functional unit. The system hardware configuration is described. Daily and yearly totals for the amount of CT, MR, and US images generated in our department are provided. The software modules for the system operation are presented. We also report on our previous experience with a MR/US PACS module for a single radiology section.

Computer networks for Picture Archiving and Communications. Systems (PACS) have evolved over the last few years. Twisted pairs and coaxial cable networks have been used for image transfer at low data rates. Second generation PACS networks use fiber optic communications at speeds up to 140 Mbps. In this paper, the need for integrated voice, data, and image communications and network rates over 200 Mbps is presented. A high speed fiber optic network for integrated PACS services has been design and simulated in the Computer Engineering Research Laboratory at the University of Arizona. This paper summarizes the characteristics and protocols of this network. The network represents a high performance application for local PACS environments and its implementation is technically feasible now.

History: Our experience with Medical Image Management Systems (MIMS, also called PACS) began in 1982 with the creation of a digital subtraction angiography (DSA) unit. This DSA system was built utilizing a DeAnza image array processor with boards fabricated by our staff to interface with a Siemens angiography room. Because of the need to transmit and eventually store very large image files on a remote computer, we designed and fabricated a point-to-point fiber-optic link [82-ARENA. This device was later marketed by Canoga Systems and was an important contribution to the design of commercial fiber-optic networks. Recognizing the importance of a versatile Radiology Information System (RIS) and its critical inter-relationship to a MIMS, some very early work on RIS design was carried out [79-ARENA, [79.B-AREN ], [84-AREN ].

Picture Archive and Communications Systems (PACS) represent a significant long term capital investment for radiology departments and hospitals. Many radiology departments want to acquire this new imaging technology, but they are still concerned about the cost of these systems. While a few studies have tried to quantify the costs and benefits of PACS, these studies have been limited in scope. The University of Washington is evaluating a Digital Imaging Network System (DINS) and PACS for the U.S. Army Medical Research and Development Command. Part of this evaluation includes developing a comprehensive cost model of PACS for one of the military's large health care facilities (a 400 bed hospital). The paper summarizes the methodology and multi-layered spreadsheet model developed at the University to forecast the costs and potential cost savings this health care facility might accrue if a hospital wide PACS is installed and film is eliminated. It also discusses the many important assumptions made in the model. A sensitivity analysis of the model is also presented. The model indicates that keeping PACS maintenance costs down is particularly critical to the cost effectiveness of PACS. That is, the film cost savings attributed to PACS can be largely offset by PACS equipment maintenance cost. The cost effectiveness of PACS will also hinge upon whether a number of intangible benefits, such as referring physician and support staff productivity gains can be attributed to PACS. This model also suggests that the pay back period for a hospital wide PACS will vary significantly dependent upon the mix of tangible versus intangible cost savings incorporated into the modeling process.

PACS has represented an unfulfilled vision for many years, although many groups have now begun to implement and test system elements. To realize a complete PACS requires a dedicated effort between real users of PACS and technology providers. This paper presents the details of a joint effort between the Mallinckrodt Institute of Radiology, Washington University in St. Louis, and Digital Equipment. Corporation to establish a PACS Testbed with the goals of identifying the key deficiencies in the critical component technologies, and understanding the significant system integration issues of an operational PACS. This Imaging Testhed will strive to advance the "state-of-the-art" in the areas of network systems, large file database storage systems, high-resolution grey scale display workstations, integrated heterogeneous network applications, and international standards.

This paper describes installing the first two portions of a fiber-optic based image-transmission network. The network topology is a star. The system is controlled by logical circuit switching of broadcast signals. High-speed operation is attained by two means: First, the signalling speed of the network is 144 Megabits/second. Second, wavelength multiplexing is used to separate the control signals and image-transmission signals. This paper describes the installation and integration of two arms of the network star. The arms include: the star coupler (SC), two network interface units (NIU), two interface translation units (NIU), and two units of imaging equipment (IE). Development of the system components was done separately at the Department of Radiology, University of Arizona (UofA) in the U.S.A. and at the Toshiba Medical Imaging Laboratories in Japan. The SC and NIU's were constructed by Toshiba while the ITU's and IE's were constructed by the UofA. The system integration was done at the University of Arizona by a team with members from both locations.

In this paper, we present a simulation study of a browsing system involving radiological image servers. The proposed IEEE 802.6 DQDB MAN standard is designated as the computer network to transfer radiological images from file servers to medical workstations, and to simultaneously support real time voice communications. Storage and transmission of original raster scanned images and images compressed according to pyramid data structures are considered. Different types of browsing as well as various image sizes and bit rates in the DQDB MAN are also compared. The elapsed time, measured from the time an image request is issued until the image is displayed on the monitor, is the parameter considered to evaluate the system performance. Simulation results show that image browsing can be supported by the DQDB MAN.

This paper describes a digital teleradiology system developed by the DuPont Corporation and some of the psychophysical and clinical evaluations we have conducted on the system in the last two years.

Teleradiology systems were used for after-hours consultation and weekend consultation in the practice of radiology. Six different systems were evaluated over a five year period. The current paper deals with an overview of the project which took place over a period of five years with approximately 20,000 images transmitted. Images obtained with the use of computerized axial tomographic equipment was the primary application.

Document architecture is a fundamental element in the design architecture of Picture Archiving Communication Systems. This paper gives an overview of a multimedia document architecture in terms of logical and layout structures and describes a method for organizing and modeling multimedia diagnostic reports. The proposed document model is based on the Office Document Architecture (ODA) ISO standard with additional enhancements to satisfy the functionality required of future interworking PACS.

The diagnostic workstation serves as the radiologist's interface to the PACS and is intended to ultimately supersede the film alternator as the radiologist's principal tool for image reviewing, reporting, and consulting. At the University of Washington, we have developed a personal computer-based, single-screen, radiological imaging workstation and two supporting user interfaces. The first interface, RadUI I, used pictorial directories to offer the user a quick visual overview of a single patient study. In addition, it made available a variety of image processing functions through a mouse-based, menu-driven user interface. A second user interface was motivated by our belief that the display of images, rather than image processing, is the fundamental issue in workstation design for radiological viewing. The RadUI II user interface supports the tiled display of up to 4 independent, virtual display monitors on a single physical screen. Both interfaces are based on a hybrid alternator-filmstrip conceptual model. This paper describes the two user interfaces and reports the results of an evaluation designed to study the effectiveness of the alternator-filmstrip model.

The radiology department of the future will be dramatically different in its operations because of the rapid development of digital radiography devices and image management technology. The introduction of new diagnostic devices such as MRI, CT, and SPECT has given powerful tools to diagnosticians, without changing the basic operational mode of radiology service. Recent advances in computed radiography (CR) and image management and communication systems (IMACS) have made it possible to move a significant portion of radiology service toward a filmless environment. It is increasingly clear that within the next several years a fully automated radiology service, with only a limited use of films, can be achieved. In such an environment the operations of a radiology service may be significantly different from those of a film-based service.

Advances in fiber optics and ISDN communications technologies will enhance the communications capabilities of PACS. These enhancements will provide PACS users with continued fast access to exam data while supporting the future growth of PACS. For example, PACS will be able to handle higher resolution images, larger numbers of exams, larger numbers of users, and additional applications. This paper provides a brief tutorial on PACS networking and the impact of the new fiber optic and ISDN technologies on the evolving PACS network. The network model, topology, traffic, and performance requirements will be described, and topics such as standards, cost, and flexibility will be addressed.

A complete image management and communications system has been installed at Georgetown University Hospital (GUH). The network is based on the A T & T CommView^® System. In the Neuroradiology Division, this comprehensive network supports a multiscreen workstation with access to multiple imaging modalities such as CT and MRI from both the hospital and a remote imaging center. In addition, the radiologist can access these images from various workstations located throughout the hospital as well as from remote sites such as the home. Among the radiology services supported by the network, neuroradiology has the greatest need for such a system with extensive daily requirements involving the remote imaging center and on-line consultation around the clock. By providing neuroradiology with all available communication links, the radiologist can monitor, diagnose, and consult. The remote site has a subsystem capable of acquiring images and transmitting them over a high speed T1 data circuit. The GUH neuroradiologist can view these images on the neuro workstation or any of the workstations available in the Hospital. Fast and easy access to the images allows a radiologist to monitor multiple examinations as well as to utilize the workstation for diagnosis. To provide the neuroradiologist quick access to images at all times, a PC-based Results Viewing Station (RVS) has been placed in a doctor's home. Images may be sent to the RVS, or the user may request images from the central database at the hospital. Images can be viewed at home either as they are transmitted, or following transfer of a whole study. The efficiency and effectiveness of the system's capabilities with special regard to remote and teleradiology (RVS) operations have been studied for the neuroradiology service. This paper will discuss the current clinical acceptance and use, problems in implementation, and ways these difficulties are being surmounted.

A largely overlooked issue in the Picture Archiving and Communication Systems (PACS) area is the tremendous amount of site planning activity required to install such a system in an existing medical center. Present PACS equipment requires significant hospital real estate, specialized electrical power, cabling, and environmental controls to operate properly. Marshaling the hospital resources necessary to install PACS equipment requires many different players. The site preparation costs are nontrivial and usually include a number of hidden expenses. This paper summarizes the experience of the University of Washington Department of Radiology in installing an extensive digital imaging network (DIN) and PACS throughout the Department and several clinics in the hospital. The major logistical problems encountered at the University are discussed, a few recommendations are made, and the installation costs are documented. Overall, the University's site preparation costs equalled about seven percent (7%) of the total PACS equipment expenditure at the site.

The communication performance aspects of a PACS in the department of Radiology in a hospital of reasonable size were investigated by means of an analytic model. Assuming a realistic load in number of newly generated images and reporting by radiologists, it turns out that the system can be designed in such a manner that a communication speed of 8M bytes/sec will be sufficient. In such a system much attention must be given to pipeline the operation of many processes. This applies to the design of the ACR-NEMA communication layers and to the fetching and storing of messages on disks (application layer). In turn, operations in the application layer can only run in parallel if the short Commands will be given transmission priority in the communication net, in order to set up the fetching of a message from disk as early as possible. Parallel operation also requires the provision of buffer memories at the workstations and acquisition devices.

The interface of radiology information systems (RIS) to image management and communications systems (IMACS), also known as picture archiving and communication systems (PACS), is necessary if they are to become clinically accepted. At Georgetown, we have started development of an interface between our MIIS based RIS and our AT&T CommView system. This interface is based on a one-way flow of information, with data transferring from the RIS to the PACS only. The interface is invisible to the user and requires no special action by the user to facilitate data transfer.

The digital imaging network (DIN) project is a federally funded demonstration and evaluation project which studies the usefulness of emerging management technologies for radiological images. The primary emphasis of the university-based efforts is in the clinical operation of the entire network. The project is interested in the evaluating the concept rather than any particular product.

A comprehensive image management and communication (IMAC) network has been installed at Georgetown University Hospital for an extensive clinical evaluation. The network is based on the AT&T CommView system and it includes interfaces to 12 imaging devices, 15 workstations (inside and outside of the radiology department), a teleradiology link to an imaging center, an optical jukebox and a number of advanced image display and processing systems such as Sun workstations, PIXAR, and PIXEL. Details of network configuration and its role in the evaluation project are discussed.

The Department of Electrical Engineering at the University of Ottawa has developed a multimedia medical communication facility. The purpose of this system is to facilitate communication among clinicians and radiologists. More timely information should be available to the clinician using this system, as well as providing a convenient consultation mechanism . The multimedia workstation must also be of sufficient quality that the radiologist's report can be generated from the workstation without the actual x-ray films. From the physician's standpoint, the system must be easy to learn, easy to use, and have the appropriate functionality to accomodate diagnosis and report generation. This paper discusses the results of an evaluation of the imaging facilities from this perspective.

One area of PACS research at the Georgetown University Medical Center, Department of Radiology, relates to the user interface of the PACS workstations. Other research areas include network simulation, data compression, digital radiology, image processing, and teaching workstation development. The major focus of this paper is a presentation of a PACS user interface issues, including what should be considered when designing a man- machine interface for a PACS workstation. A brief discussion of PACS requirements is presented, followed by the advantages and disadvantages of various interaction techniques and devices. Next, a set of requirements for a PACS user interface prototyping environment is specified. Finally, the hardware and software components of the Georgetown PACS research environment are presented. Both those components which are applicable to user interface prototyping and those used for other PACS-related projects are discussed.

At the University of Ottawa Medical Communications Research Centre we are looking at the PACS opportunity from the perspective of improving the communication of Radiological information (images and diagnostic reports) between the Radiologist and the client Physicians. To that end, we have developed an experimental multimedia communication system (called IRIS) which supports the capture and distribution of digitized X-Ray images and voice reports as integrated "electronic" patient folders. It also supports online consultation between the Radiologist and Physician by providing a multimedia "shared space" through synchronized workstation operation. In order to assess the clinical value and acceptance of these service concepts, the system is being deployed in the Ottawa Civic Hospital for a series of clinical trials beginning in early 1989. This paper describes the architecture and operation of the IRIS system, and briefly discusses the first clinical trial -an Emergency/Radiology link.

Digital image archives are central to the operation of PACS systems for medical images. Current technology that supports the terabyte requirements of today's radiology departments include magnetic tape, write once optical disk (WORM), erasable optical disk, and optical tape. We review current storage technology and place it in the context of current database requirements for radiological digital imaging.

A 10-12 bit gray scale is provided in commercial laser film digitizers. The true contrast resolution on the digitized image within 200 microns square is limited by both the quantum mottle and instrumentation noise. In this paper, we investigate that mean value, standard deviation, and adjacent pixel correlation coefficient on a calibrated step wedge film with two laser digitizers. The results were disappointing. On the output side, the brightness of different grey levels from a clinical monitor was measured with a narrow angle luminance probe. In addition, the imlipcations for teleradiology applications are discussed in this paper.

One of the goals of the Dutch PACS-project (a collaborative effort of the Utrecht University Hospital (UUH), Philips Medical Systems and BAZIS Central Development and Support Group Hospital Information Systems) is to perform a clinical evaluation and equipment testing at the UUH. In earlier publications a setup for this clinical evaluation is presented. A target ward, a subdivision of the department of Internal Medicine (IM) with a total number of 15 beds, is selected to be treated fully digital. Two other wards, similar to the target ward, were not digitally treated and served as control group. To exploit the efficiency potential of PACS, a comparison will be made between current procedures and PACS procedures. From February 1st until June 1st 1988, a detailed timing and logistic study of examinations and folder tracking has been carried out for all patients of the 3 wards of Internal Medicine, mentioned above. The first part of the study regarding the conventional distribution process is now completed. For this purpose time-registration forms were added to the patients application-forms for radiological examinations. These forms were required to be filled in by all personnel involved (receptionist, archive, technologist, radiologist, etc.). A total number of 500 usable time-registration forms were gathered. One of the results, indicated that the mean time interval for examinations to cover the total conventional distribu-tion process comes to 53.9 hours. An analysis of the number of disturbances in the conventional distribution process indicated that in 24.1 % of all examinations, some exponent was lacking (missing, loaned, delayed, etc.). These disturbances are analyzed in relation to time-intervals. Data were analyzed statistically by means of the SPSS statistical package. This paper describes and explains in detail: - frequencies and means of the total distribution time and different parts of the current distribution process - total number of disturbances in the current distribution process - disturbances analyzed in relation to time intervals The three wards were statistically not significantly different, so they can be compared with each other. This analysis shows clearly the possibilities for efficiency of the PACS distribution process and the consequences of this innovation for the organization and organizational communication.

We have investigated the use of a simple expert system to generate a PACS configuration from a simple set of specifications. This paper will describe the initial expert system and its rules and the simulation display tool. While there have been several other projects which have simulated PACS systems, none have tried to automatically generate system configurations.

Establishing control of the image data flow between a computed radiography system and a host computer presents many problems in a centralized PACS system design. Dedicating the main computer of the PACS system to service on-line acquisition devices will soon overload the system, especially when multiple image acquisition units are present. At UCLA, our approach is to distribute the task of buffering the input image data to a smart file server. Using a low-cost microcomputer as our image file server, we can reduce the burden to the central PACS computer when an image is to be acquired. This paper describes the design of the smart file server for a PCR/SP computed radiography system.

An iterative prototyping approach has been used in the development of requirements for a new user interface for the display workstation in the CommView® system product line. This approach involves many steps, including development of the preliminary concept, validation and ranking of ideas within that concept, prototyping, evaluating, and revising. We describe in this paper the process undertaken to design and evaluate the new user interface. Staff at Abbott Northwestern Hospital, Bowman Gray/Baptist Hospital Medical Center, Duke University Medical Center, Georgetown University Medical Center and Robert Wood Johnson University Hospital participated in various aspects of the study. The subject population included radiologists, residents, technologists and staff physicians from several areas in the hospitals. Subjects participated in in-depth interviews, answered questionnaires, and performed specific tasks, to aid our development process. We feel this method has resulted in a product that will achieve a high level of customer satisfaction, developed in less time than a traditional approach. Some of the reasons we believe in the value of this approach are: • Users may not be able to describe their needs in terms that designers are expecting, leading to misinterpretation; • Users may not be able to choose between options without seeing them; • Users needs and choices evolve with experience; • Users true choices and needs may not seem logical to one not performing those tasks (i.e., the designers).

The existence of a link between a HIS, incorporating a RIS, and a PACS is one of the most indispensible prerequisites to perform a clinical evaluation study of a PACS in practice. The two main reasons for realizing a IIIS-PACS coupling are concerned with 1) proper image management e.g. in order to achieve acceptable waiting times and 2) the user need of having all patient relevant data available in the same workstation. A first phase coupling between an existing HIS and a just installed PACS has been realized by BAZIS in the Utrecht University Hospital in the Netherlands. These activities were carried out as part of the Dutch PACS Project, a collaboration in the PACS field of BAZIS, the Utrecht University Hospital (AZU) and Philips Medical Systems. For the first phase only HIS data of the in-patients of one particular ward of Internal Medicine were transmitted using a one-way dataflow (from. HIS to PACS). In phase two the coupling will he hi-directional and more departments will be part of the experiment. In phase three of the coupling project a more general. HIS-PACS interface will be developed, independent of the HIS and PACS manufacturers. In April 1988 the first public data exchange (patient data, order data, reports) between the BAZIS/ZIS and Philips/MARCOM system has been demonstrated. This was the first working example of a HIS-PACS coupling worldwide to our knowledge. This paper gives an overview of the experimental set-up, the demonstration during the 6th EuroPACS meeting , the results and problems encountered thusfar and statistical data from daily practice.

The most critical aspect of a radiologist's work is the communication of his findings to the attending physician responsible for the patient's care. This is also the part of the process that is least well organized and the most subject to failure. At the University of Ottawa Medical Communications Research Centre we are investigating technical means to improve communications between radiologists and attending physicians. We first introduce the radiology communication service problem and show why it is essentially a multimedia communication problem. We then briefly describe a multimedia communication system designed and implemented by our research team. The multimedia system consists of several workstations linked by the Hospital's LAN. Each physician workstation comprises a Compaq 386/20 Mhertz microcomputer with 16 Mbytes of RAM, a 500 Mbyte image disk, an image memory which drives a 1000 line monochrome monitor. The images are digitized using a Konica laser-based film digitizer (2430 by 2000 10-bit pixels for a standard chest radiograph). The multimedia file server manager station is built around a PC-AT compatible with a Northern Telecom MERIDIAN® SL-1ST digital PBX and a Meridian Mail digital voice messaging system. This last device is used to store voice data and is linked via the PBX to the workstations' digital telephones. A SYTEK® 6000 local area network (LAN) links all workstations to the file server. All data, image and graphic information is transmitted via this network, while the twisted pair connections linking the digital PBX to the telephone sets are used for transmitting voice data. Finally, we provide details of an in-hospital trial linking the Department of Radiological Sciences and the Emergency Department at the Ottawa Civic Hospital, a 950 bed tertiary care teaching hospital.

A few of us here can look back ten years or more of having observed or even having participated in the PACS development. Depending on our point of view, little or much progress has been made since 1978. In the following, the authors will sketch a log book of an industrial developer and vendor, i.e., Philips Medical Systems. Some comments will turn out to be specifically applicable to the Philips Concern but the majority is probably also true for other major X-ray manufacturing companies. There are certain ground rules which have to be observed if a company wants to prove itself in a market as ambitious and complex as the medical market and, in particular, the PACS market. These ground rules will be either explicitly mentioned or are implied for the attentive listener. The following aspects will be considered:

Network performance analysis is an important part of PACS implementation. An understanding of the operational PACS - is necessary for the development of a realistic computer model of the system. Subsequent simulations of the model will help locate under and over utilized nodes and predict the effect of changes to the network. At Georgetown, we have started a model of our AT&T PACS, by first studying the acquisition aspect of the PACS in detail and developing a computer model using PAW simulation software.

To be successful, a Picture Archiving Communication System (PACS) workstation must have a user interface which is easy for radiologists to use. This paper describes the iterative design methodology used in the Integrated Radiology Information System (IRIS) project to achieve an easy to use system. Iterative design involves the radiologists in the design effort throughout the project. Iterative user interface design requires a number of activities: goal setting, identification of important user interface issues, prototyping, structured user testing, evaluation of user-test results against the goals, and further prototyping if the goals are not met. These stages are described in detail with examples taken from the IRIS project.

The impact of PACS on the design of various hospital departments is examined. The functionality requirements imposed by these departments on PACS components and on the Hospital Information System are discussed.

The use of a Digital Radiography system (DR) with a digital interface to a PACS, greatly increases the potential of the system for effective communications, since it makes the majority of the cases done in the radiology department available without the need to digitize film. The new all digital, fully integrated system, consisting of a DR system and a PACS has been installed in the Duke University Medical Center Radiology Department. The digital radiography system, Philips Computed Radiography (PCR), transmits images to the PACS (CommView® ) via a PACS Interface Processor (PIP). The questions and problems associated with the integrated system are now being investigated.

Dearchival of image data in the form of x-ray film provides a major challenge for radiology departments. In highly active referral environments such as tertiary care hospitals, patients may be referred to multiple clinical subspecialists within a very short time. Each clinical subspecialist frequently requires diagnostic image data to complete the diagnosis. This need for image access often interferes with the normal process of film handling and interpretation, subsequently reducing the efficiency of the department. The concept of creating a local image database on individual nursing stations utilizing the AT&T CommView Results Viewing Station (RVS) is being evaluated. Initial physician acceptance has been favorable. Objective measurements of operational productivity enhancements are in progress.

A PACS must provide a user interface which is acceptable to all potential users of the system. Observations and interviews have been conducted with six radiology services at the Georgetown University Medical Center, Department of Radiology, in order to evaluate user interface requirements for a PACS system. Based on these observations, a conceptual model of radiology has been developed. These discussions have also revealed some significant differences in the user interface requirements between the various services. Several underlying factors have been identified which may be used as initial predictors of individual user interface styles. A user model has been developed which incorporates these factors into the specification of a tailored PACS user interface.

One clinical environment which can immediately benefit from the implementation of a radiologic PACS is the intensive care unit (ICU). Our previous study has demonstrated the feasibility and timeliness of routine image transmission to an ICU. In anticipation of future expansion of this service, we have investigated two different models for a hospital-wide ICU PACS. These models included a centralized and a distributed processing PACS configuration. Their comparison indicated that although the distributed model offers some major advantages over the centralized model, the latter may hold a rightful place in the inter-departmental service, especially if the cost issue is a critical factor.

In building an integrated radiological information system (IRIS) and preparing for field trials at a large hospital, it was necessary to consider the working conventions and needs of radiologists and clinicians. It was also necessary to analyse the current practices involved in the requesting, reading, and reporting of x-rays. Because IRIS is designed to be a mechanism for effective consultations, an understanding of the conventions and nature of x-ray related consultations was also needed. In order to gain this understanding, two observational studies were conducted. The first addressed issues of x-ray related behavior in the emergency department, and the second assessed the task requirements of radiologists when reporting x-rays or consulting with other physicians about x-rays. These findings had major implications for the design of IRIS and for conducting the field trials.

A concise overview is presented on the results of the total IMAGIS (IMAGe Information System) research as carried out by BAZIS. This paper is intended as a continuation of the IMAGIS presentation at the 'Dutch PACS session' during the SPIE Medical Imaging H Conference 1988. That session was jointly organized by the Utrecht University Hospital (AZU), BAZIS and Philips Medical Systems, the partners within the Dutch PACS project. The HIS-PACS coupling/integration project has resulted in a HIS-PACS coupling that is used to transfer data from the BAZIS HIS to the prototype Philips PACS. The modelling and simulation project has resulted in a modelling and simulation package (MIRACLES), which has been used to suggest performance improvements of existing and future PACS's. To support the technology assessment project a PC program has been developed that calculates the financial consequences of the introduction of a PACS (CAPACITY). The diagnostic image quality evaluation project has provided research protocols and a software package (FEASIBLE) that have been used as an aid to execute observer performance studies. A software package (FRACTALS) has been developed so that a standard computer (113M RT PC) can be used as a simple image workstation for radiological research. Furthermore, a number of general issues concerning the development, the acceptance and the introduction of PACS are discussed.

Picture archiving and communication systems (PACS) are now clinically available in limited radiologic applications. The benefits, acceptability, and reliablity of these systems have thus far been mainly speculative and anecdotal. This paper discusses the evaluation of a PACS module implemented in the pediatric radiology section of a 700-bed teaching hospital. The PACS manages all pediatric inpatient images including conventional x-rays and contrast studies (obtained with a computed radiography system), magnetic resonance images, and relevant ultrasound images. A six-monitor workstation is available for image review.

The diagnostic image quality of a Philips' Picture Archiving and Communications Subsystem (PACS) was investigated by means of a Receiver Operating Characteristic study as a contribution to the current Dutch PACS research project. Conventional films and laser- and videodigitized images were compared for diagnostic image quality. A test series of 68 posterior-anterior (PA) screen film roentgenograms of a 3M chest phantom with superimposed simulated interstitial pattern disease were produced. The conventional test films were dig;tized using a video- and laser digitizer. The series of conventional and laser- and videodigitized images were viewed by four experienced radiologists. The digitized images were displayed on a 60 Hz interlaced monitor with 1024 lines. The relative ranking in terms of decreasing ROC areas was respectively conventional, laser- and videodigitized images.

At BAZIS modelling and simulation are extensively used to investigate and predict the performance of (future) image information systems. This paper will describe our practical experiences as well as the data obtained with modelling and simulation during the past 4 years. It has proven to be very useful to use these techniques as decision support tool when developing PAC systems. Apart from performance prediction, modelling and simulation have also been used for hypothesis testing ("what-if-questions"). Besides, modelling and simulation have resulted in detailed insight into practical problems which might arise when introducing PACS. In order to facilitate the construction of simulation models of image information systems, we developed the modelling environment MIRACLES. The paper gives an overview of the (prototypical) PAC systems which have been simulated with MIRACLES. Finally, in this paper the future activities concerning the modelling and simulation of image information systems and their clinical environments will be discussed.

In formulating an Implementation strategy for digital medical imaging, three interrelated thrusts have emerged for the Defense Medical Establishment. These thrusts- (1) Totally filmless medical imaging on the battlefield, (2) Teleradiology, and (3) DIN/PACS for peacetime military health care- have significant implications in their fully developed form as resource savers and quality improvers for the unique aspects of military health care .

PACS's (Picture Archiving and Communication Systems) have matured to the commercial product stage. These systems introduce capabilities to the radiology field which can only be delivered by digital computer technology. PACS operates using workstations that are designed and configured for the tasks at hand. Various PACS configurations are possible: multi-modality, single-modality, and systems with limited focus, such as picture archiving or teleradiology. PACS installations can cost from the neighborhood of $100,000 to many times that amount. The overall cost depends on the size of the system, the characteristics of the central computer installation and those of the distributed workstations.

The management of radiological images and associated information has always been a difficult task, but technologies are being integrated to address this problem. One of the many acronyms for this technology is DIN, digital imaging network. One of the most significant obstacles to the implementation of this technology is the high cost. While cost has not prevented the radiology community from assimilating other new technologies, the major difference between previously assimilated technologies and DIN technology is that DIN produces no directly billable service and the impact on patient care is less obvious. What is the cost associated with the existing problems in radiological image management? What is the cost of solving these problems using DIN technology? Can we afford this technology? The CAMDIN model addresses these issues.

The operational analysis project at UNC is continuing, with a major goal directed at prospective analysis of the effectiveness of technological capabilities and potential cost effects of PACS on the operation. We have now defined the major operational variables we believe will provide information to evaluate PACS technology. These operational variables, the modeling methodology and the methods of data collection are described. Our approach to statistical analysis of diagnostic quality is also briefly discussed.

Small computers incorporating hard disc memory, multiple high resolution monitors and the small computer systems interface (SCSI) can be used for low-cost filmless radiography. A system has been constructed which can perform all of the functions required of a small clinic or field hospital including scheduling, reporting, image acquisition and display, image annotation, image storage and transmission, and control of peripheral devices. The peripheral devices include an optical card reader/writer, an optical disc reader/writer, a SCSI to DIN/PACS port, an Ethernet port and a SCSI to a long distance telephone/computer port, the Integrated Services Digital Network (ISDN) port. Individual patient optical data cards may be prepared, all images and reports may be archived in a small optical disc in the computer, other image sources may be coupled to the system via the DIN/PACS port, data may be exchanged with the local DIN via the Ethernet port and with distant sites via the ISDN port. The small optical data cards, about the size of a credit card, are used for individual patient images and reports. An independent viewer may be used to display the contents of the cards. The result is a complete "filmless and paperless" medical imaging system. The system was developed on Contract DAMD17-88C-8058 with the US Army Medical Research and Development Command.

We have studied the problem of adaptive cosine transform coding, and developed an original quantization algorithm which can minimize block artifacts. There are three features in our compression technique. The first feauture is variable quantization interval. The quantization interval is determined adaptively depending on the statistical characteristics of the block. In our technique, the number of AC components which are rounded off to zero by quantization is used for the quantization parameter. The second feature is the adaptive cutoff of the high frequency components of the transform coefficients. High frequency components may be ignored adaptively in our technique. The third feature is the double quantization algorithm. Non-adaptive quantization is carried out first. The second quantization is performed adaptively.

Data stored in a PACS archive must be organized to permit efficient retrieval. The concept of a unique data object identifier (UID) permits a fundamental partitioning of the problem into a storage system indexed by UID and a database containing descriptive elements. The database serves to map user retrieval requests, expressed in terms of clinically relevant descriptive elements, into UIDs of specific data objects. Different data organization mechanisms are employed by imaging modalities, thereby making the structure of a generic PACS database difficult. One solution may be derived by analogy from film based systems. Images and other data objects may be organized, by application of modality and site specific rules, into electronic folders. Folders, in turn may be organized into a patient master folder or user defined reference folders. Such an organization provides an easily understood user access model for information stored in PACS archives. This paper presents a PACS architecture comprising an Information Management System (IMS), and Image Storage System (ISS). Entity-Relationship diagrams are presented to define a schema for the IMS database based on the folder analogy. The folder concept and its relationship to the ACR-NEMA Standard are discussed.

The advent of second-generation optical storage technology and user experience with PACS dictate a new look at PACS long-term database organization and optical archive strategy. Optical disk drives are growing faster and more reliable, disks and jukeboxes will hold more data for their size, and the cost per byte of optical disk storage will fall. Users are becoming more sophisticated in their use of PACS. Field experience has increased our knowledge of user needs and refined our models of optical archive usage. COnsidering these advances, this paper addresses such topics as centralized archives, standards and equipment compatibility, flexible archival and de-archival criteria, data reliability, backup strategies, and improved archive performance.

The authors have been using medical image filing system TOSHIBA TDIS-FILE as a teaching files archiving and viewing at University of Tokyo, Hospital, Department of Radiology. Image display on CRT was proven sufficient for the purpose of education for small groups of students, as well as residents. However, retrieval time for archived images, man-machine interface, and financial expenses are not in a satisfactory level yet. The authors also implemented flexible retrieval scheme for diagnostic codes, which has been proven sophisticated. These kinds of software utilities, as well as hardware evolution, are essential for this kind of instruments to be used as potential component of PACSystem. In our department, PACS project is being carried on. In the system, TOSHIBA AS3160 workstation (=SUN 3/160) handles all user interfaces including controls of medical image displays, examination data bases, and interface with HIS.

By the early 1990s, the U.S. Army intends to implement all-digital, filmless medical imaging throughout its deployable combat health care facilities. In addition to employing imaging devices whose outputs are comprised of digital, electronic images rather than film-based images, such a system must, at a minimum, include means for storing, displaying, and transporting the resultant images in accordance with existing military doctrine. Furthermore, compatibility with fixed medical facilities, both in the continental United States (CONUS) and elsewhere, and between military services must be assured to promote continuity of care. The aim of this undertaking is to develop an understanding of the imaging needs of the U.S. Army's deployable/combat Health Services Support (HSS) system, and address these needs in a prototype Digital Imaging Network System (DINS) (also known as Picture Archiving and Communication System (PACS) or an Image Management and Communication (IMAC) System), to be integrated and evaluated by The MITRE Corporation. The prototype will undergo iterative development by subjecting it to multiple evaluations in both the laboratory and Echelon 3 environments, by military and civilian medical, scientific, and engineering personnel. This project is part of a U.S. Army Medical Research and Development Command contract to The MITRE Corporation to assess the feasibility of filmless imaging in military health care.

Ongoing research and evaluation projects sponsored by the Army Medical Research and Development Command are leading towards filmless radiography in the combat casualty care system of the 1990s. With the elimination of film, the question arises as to the most appropriate medium for archiving and transporting x-ray images and related patient data with the wounded between facilities. This paper considers the pros and cons of the various candidate media in relation to their specifications, availability, and appropriateness under simulated combat casualty care conditions.

To evaluate whether the new technology of PACS can be cost justified, we have developed a computer model to analyze the costs of manual vs. digital PACS. The model was used to compute the costs of manual PACS for 14 imaging centers in 1988. The results were compared against the cost profile of an average US 400+ bed hospital derived from '85-'87 TMG PACS tracking studies in over 600 US institutions. The comparison showed that the model accurately predicted costs in these 14 imaging centers to be close to the US average. The total predicted cost per average examination was $21.49. The total annual costs were found to have a linear relationship to the number of examinations and the number of beds in the hospital. The model indicates the annual cost of a manual PACS at a 500 bed hospital to be $ 2 million.

A computerized Radiology Management System was used to monitor the turnaround time of radiology reports in a large university hospital. The time from patient entry into the department until the printing and distribution of the final examination report was monitored periodically for two-week time intervals. Total turnaround time was divided into four separate components. Analysis of the data enabled us to assess individual and departmental performance and thereby improve important patient service functions.

In a contribution to the Medical Imaging,II conference 1988 the design and construction of a package for PACS cost modelling was advocated. In this paper the main features of such a package are described. A first version of the package, called CAPACITY, is ready and experiences are being collected. The package produces as its main results a comparison oftotal cost differences of a PACS configuration and a film-based system as a function of time. The user has to describe the PACS configuration, the technical characteristics of the various components together with the estimated price (as a function of time). Also for the film based system the costs of the various components have to be defined by the user. The production volume of the hospital and the radiology department is input and used in the critique module that checks the input data supplied against a number of rules of thumb (to be defined by experts). At present the package is being used experimentally within our group. At the next EUROPACS meeting (March 1989) it will be demonstrated and other groups will be invited to use it. It is expected that discussions on the results of the package and analysis of the input data supplied will contribute to convergence of the widely varying opinions about the costs of PACS and the moment PACS will be cost justified (both within the radiology department and in the clinic).

Computed radiography (CR) and film digitizers have now made it possible to begin conversion of standard film archives to electronic storage of image data. Advantages of electronic storage include increased image availability; decreased image loss; electronic image enhancement and analysis; and savings in space and possible savings in personnel time required to operate the image archive.

Our institution (San Francisco VA Medical Center) is a VA pilot center for total digital imaging and PACS. Quantitative information about PACS impact on health care is limited, because no centers have done rigorous preimplementation studies. We are gathering quantitative service delivery and cost data before, during, and after stepwise implementation of computed radiography and PACS at our institution to define the impact on imaging service delivery. We designed a simple audit method using the x-ray request and time clocks to determine patient waiting time, imaging time, film use, image availability to the radiologist, matching of current with previous images, image availability to clinicians, and time to final interpretation. Our department model is a multichannel, mulitserver patient queue. Our current radiograph file is space limited, containing only one year of images; older images are kept in a remote file area in another building. In addition, there are 16 subfile areas within the Radiology Service and the medical center. Our preimplementation audit showed some long waiting times (40 minutes, average 20) and immediate retrieval of prior films in only 42% of cases, with an average retrieval time of 22 hours. Computed radiography and the optical archive have the potential to improve these figures. The audit will be ongoing and automated as implementation of PACS progresses, to measure service improvement and learning curve with the new equipment. We present the audit format and baseline preimplementation figures.

As a step toward computer assisted imagery interpretation, we are developing algorithms for computed radiography that allow a computer to recognize specific bones and joints, and to identify variations from normal in size, shape and density. In this paper we report on our approach to model-based computer recognition of hands in radiographs. First, image processing hypotheses of the imaged bones. Multiple hypotheses of the size and orientation of the imaged anatomy are matched against stored 3D models fof the relevant bones, obtained from statistically valid populations studies. Probabilities of the hypotheses are accrued using Bayesian inference techniques whose evaluation is guided by the structure of the hand model and the observed image-derived evidence such as anti-parallel edges, local contrast, etc. High probability matches between the hand model and the image data can cue additional image processing-based ssearch for bones, joints and soft-tissue to confirm hypotheses of the location of the imaged hand. At this point multipule disease detection techniques, automated bone age identification, etc. can be employed.

This document describes the use of optical delay lines for data storage and a series of experiments to demonstrate this idea. The idea is based on constructing an optical version of a mercury delay line - such as those used in the 1950s.

The evaluation of the Dutch PACS project comprises amongst other studies a comparison of the direct financial costs of the present conventional system and the PACSystem. This paper shows that the invest-ment costs of a PACSystem are presently substantially higher than of the present system. These higher investment costs will not be compensated for by savings in the area of materials used, floor space needed, and personnel employed. The only cost effective compensation can be found in a shortening of the length of patient stay in the hospital. Preliminary results of the ongoing experiment suggest that compensation is achieved by an average decrease of one third day. Conditions under which a PACSystem can be more cost effective are being discussed.

The present phase of the clinical evaluation within the Dutch PACS project mainly focuses on the development and evaluation of a PACSystem for a few departments in the Utrecht University hospital (UUH). A report on the first clinical experiences and a detailed cost/savings analysis of the PACSystem in the UUH are presented elsewhere. However, an assessment of the wider fmancial and organizational implications for hospitals and for the health sector is also needed. To this end a model for (financial) cost assessment of PACSystems is being developed by BAZIS. Learning from the actual pilot implementation in UUH we realized that general Technology Assessment (TA) also calls for an extra-polation of the medical and organizational effects. After a short excursion into the various approaches towards TA, this paper discusses the (inter) organizational dimensions relevant to the development of the necessary exttapolationmodels.

Picture Archiving and Communications Systems are sophisticated computer systems designed to store and display medical images. It has been suggested that these systems may be more cost effective than film in practicing radiology because of better access to images by multiple users, better integration of information from multiple studies, and more rapid delivery of diagnostic reports to clinical physicians. Development of these systems is in its early stages and questions have arisen as to what functions an image management system should have and how it should interact with stand alone radiology information systems (RIS) and hospital information systems (HIS). Over the past decade, computer systems have undergone gradual reorganization from highly centralized hospital (or medical center) information systems toward decentralized departmental systems. The former systems often provided global functions but did not have the flexibility to meet the needs of individual departments. Some individual departmental systems were developed to meet specific requirements, but they were often unable to communicate with other systems within a medical center. Today many RIS's have the ability to provide departmental management tools, as well as to communicate with external systems. In order to achieve a filmless environment, PACS and HIS must be integrated with RIS systems to provide the level of information currently available. An additional goal of these systems is to automate operations and reduce the workload for technologists, admission personnel, file room staff, and radiologists.