The management of radiographic film is based on a centralized, sequential management strategy using individual patient film jackets as an archiving medium. This centralized manual management strategy has evolved because x-ray film has been the sole record of a patient's image information. However, the operation of any film file room always faces significant problems. Storage space for patient film jackets is rarely adequate. The demand for patient film jackets is great particularly during the first part of a patient's hospitalization. The access and utilization of the film jacket by multiple clinicians and hospital services can rarely be adequately met, Film jackets are often unavailable due to misfiling or capture in private collections of interesting cases. Ultimately, the successful operation of any film file room depends on the dedication and training of an adequate number of film file room staff, the cooperation of the clinical users and strict enforcement of check-in--check-out protocols.

Considerations of totally electronic picture archiving system (PACS) often neglect the fact that every radiology practice currently has some system for storing and retrieving images and related alphanumeric data. Although these systems are usually manual, many departments now use on-line computers to help manage film flow. In either event, the creation of electronic PACS can be viewed as a classic data processing problem of automating an existing system, and the conversion should proceed through the usual steps of documenting the existing system in detail, and conducting feasibility studies and cost-benefit analyses. Documenting current systems should be facilitated by computer-assisted PACS - particularly documenting transaction volumes which can be provided as a by-product of radiology information management systems. Similarly cost-benefit analysis should be facilitated, although the cost/benefit ratio may be less favorable when comparing automated to computer-assisted PACS. Finally, information management features such as those provided by current on-line radiology systems provide a framework necessary to realize the full benefits of automated PACS.

The radiologist serves as a consultant to other physicians in the practice of clinical medicine; the image obtained and the reported interpretation of that image represent the service rendered and are therefore of major importance (medically, legally and economically) to the radiologist. Because many radiology departments are organized along subspecialty lines or (in the case of a single department serving several institutions) along combined institutional and subspecialty lines, many patients may undergo diagnostic evaluation sequences in which several studies are performed and multiple simultaneous consultations may result. In the past, the lack of availability of multiple copies of the study (for multiple interested parties) has prevented the effective tailoring of subsequent examinations until the prior exam results were available; the advent of digital networks for PACS may result in a significant change in this procedure and, accordingly, in the pattern of interpretation, internal referral and organization of radiology departments. In addition, since clinicians may have access to studies directly and, possibly, prior to official interpretation, the nature of the relationship between the clinician and the radiologist may be altered by PACS.

Radiology Department Integration is challenged on the prediction that such integration demonstrate meaningful clinical and economic benefit. Environmental factors and their impact are examined including applications, economic, and technical aspects. Discussion focuses on key issues as they relate to demonstrating the justification for integration. Systems analysis techniques are proposed as a means of collecting sufficient quantitative data to properly assess clinical and economic impact. Results of preliminary Radiology Department Systems analysis and preliminary conclusions are discussed.

For more than thirteen years, the Missouri Automated Radiology System has served as the information management system for the Department of Radiology at the University of Missouri-Columbia. MARS has evolved into a standalone modular system touching upon all aspects of radiology management. Written in MUMPS, the software system is flexible and has been consistently upgraded so that it is currently "state-of-the-art". Many management lessons are discussed here. Perhaps most important of all, the radiology information system has envolved in into an instrument of policy. Whoever controls the system controls the department. There still exists a tug of war between the advocates of the central computer system and the users of the standalone system. With the advent of the image management system, which clearly is in the domain of radiology, it is apparent that proper function of the department requires a marriage of the information management system and the image, management system. Therefore, it is imperative that these highly professional activities be based within the department of radiology.

"ODS" and "PACS" are two acronyms that seem to be made for each other. Some of the strongest holdouts for magnetic storage will concede--at least privately--that Optical Data Storage (ODS) has a role in digital medical Picture Archiving and Communication Systems (PACS). This long-awaited technology is now a reality, bringing with it a family of ODS product configurations. For mass archival memory, we see large optical disks and long reels of sealed tapes. For compact archival memory devices, we see wallet-sized cards, small optical disks, sealed optical tape microcassettes, and adhesive labels containing laser recordable tape. The potential roles for some of these optical storage products are beginning to take form. This paper presents an overview of the Drexon family of optical storage products, with emphasis on product configurations, cost comparisons, and medical imagery apolications.

Instrumentation magnetic tape recording systems have begun to appear which use digital recording techniques in place of frequency modulated (F.M.) analog methods. This is a direct result of the instrumentation users need for storing data with greater fidelity and with better interfacing capabilities to digital computer based signal analyzers. These new recorders, based on conventional longitudinal recording geometries, are currently in production and can be used for storing digital medical images. Using standard instrumentation magnetic tape, high density digital recorders can store 35,000 images (512X512X8) at 60 images/sec rate. Rotary helical scan recorders based on industrial 1/2 inch video or 1 inch broadcast video machines have been demonstrated with capacities of up to 300,000 images at 120 images/ sec rates. It is apparent that existing high density digital magnetic tape recorders can store large numbers of images generated in digital radiography at reasonable cost, and using proven reusable recording media, several recording systems are described which can make use of this capability.

A stationary card optical data storage device is described. Digital x-ray images were stored and retrieved at high data rates. Average density can be approximately 3 x 10⁸ bits per in².A card has certain advantages for medical data archiving.

Fiber Optics is a well-defined technology now. Much of the effort has been limited to communication applications, e.g., many telephone companies, such as AT&T, GTE, etc., are installing optical fibers for long or short haul communications. However, interest in fiber optics imaging system has been growing since the first light-focusing fiber rod with gradient index profile was made with glass materials in late 1968.^{1 2} The major characteristic of the gradient index (GRIN) fiber is that each single fiber rod is equivalent to a conventional lens and can be used to transmit and focus an image. Combined with advantages of high flexibility and miniature size, this kind of fiber - with its lens-like behavior -has begun to be widely used in medical imaging systems.

A workable PACS system consists of several discrete assemblies which are linked together by data communication links for terminal and image data input and output. Two phases of development of a PACS system can be identified and labeled: emulation and enhancement. The latter is characterized by the observation "you do that so well, can you also do this?" Emulation, on the other hand, assures an accepted and working system. One of the key components of a PACS system is the emulation of the classical and ubiquitous "lightbox". This paper presents an emulation of the lightbox using a highspeed digital disk and video display on multiple high resolution monitors capable of 1024 x 1024 or 512 x 512 resolution pixel display. The emulation features are: a) loading and reloading in less than one second, b) access to between 800 and 3200 digital radiographs, and c) highspeed review forward and backward through this list of radiographs at rates up to 30 new images/second. Operation is under manual control both in rate and direction. This stresses the speed. The feature which makes this "lightbox" emulation desirable is the ability to dwell on a presently viewable display of digital radiographs with access to previous and following images in the study. One of the features desired in an emulation is that the components be "off the shelf". This means that existing hardware is used in the emulation. Software can then be generated based on known hardware. The emulation has flexibility of size: multi-image displays ranging from 2 or 3 images side by side to the piano-roll type endless display of an array with 4 images side-by-side and 3 rows visible at any one time. In addition, it has flexibility of resolution: radiographs displayable within a 256 x 256 pixel region to those displayable within a 1024 x 1024 pixel array as well as images which are much larger (4000 pixel by 4000 line chest radiograph scans) using a 1024 pixel by 1024 line window into the radiograph. The emulation of the "lightbox" finds application at both radiology review stations and at locations for consultation with attending physicians.

The history of Picture Archiving and Communication Systems (PACS) is reviewed. PACS-related activities before and after the first PACS Conference are discussed. The effect of PACS Conferences on industrial R&D programs of well established and newer companies is analyzed and documented. The challenges offered by "electronic PACS" are also discussed.

The M/NET is a local area network designed for diagnostic image distribution, storage & retrieval. The M/NET uses a broadband coaxial cable (CATV) to connect multiple remote user stations and various imaging devices. The M/NET allows instant retrieval of images and information by any of the user stations. The traditional film file room is replaced by electronic archiving center(s) thus eliminating the bottleneck of manual image handling. The M/NET standard image format is 512 x 512 pixels by 8 bits (256 levels of grey). This number of pixels per image is expandable to higher resolutions. The above format provides an economical means of electronic storage on an erasable magnetic disk. Archival storage is done on a read-after-write-once optical laser disk. With either medium, the retrieval time is minimized by computer controlled search and access. On-line image retrieval takes less than one second per image. Due to economical reasons the on-line storage does not exceed the equivalent of one week's work, or about 5,000 images. Retrieval times from the archival storage are under 10 seconds per image. The M/NET system is currently designed to handle a library of 150,000 images. Its modular design allows easy expansion to a volume of over 1.5 million. At the time of this writing, the first installation is underway in a major hospital center.

Initial components of a Picture Archive and Communication System (PACS) workbench have been installed at the Mallinckrodt Institute of Radiology (MIR) providing a set of basic "utilities" which facilitate comprehensive design studies and experiments. Each of the primary areas of picture acquisition, transport, processing, archiving and viewing are addressed by the PACS workbench.

We have implemented all digital collection, analysis, review, and archival storage of nuclear medicine data using commercially available computer hardware. In order to allow for modestly rapid retrieval of data from our archival store, an on-line directory system has been developed. The directory contains a subset of the fields which are entered into the patient's demographic data block and a pointer to the physical disk and logical location of the study. The directory is formed semi-automatically and is tolerant of operator error. The directory can be searched in under one minute for any combination of sub-fields. This directory system has been convenient and reliable for accessing our relatively small data base (about 6500 studies per year), but may be expandable to larger systems.

St. Luke's Hospital's Nuclear Medicine Department has been constructing a prototype Picture Archiving and Communication System since 1979. The purpose of this system is to interconnect existing imaging computers in the hospital with other resources such as image displays, image archives and larger mini and mainframe computers. To date, the system has evolved from simple file transfers using removable magnetic disks, through communication via magnetic tape, to the use of a high speed local area computer network. The challenge in the project has been the orderly evolution of the system, while supporting the tremendous growth of computerized nuclear medicine studies at this major cardiac center. The success in meeting this challenge has led to planning for the continued evolution of this system for the purpose of interconnecting other imaging modalities and of long-term archiving in a PACS for medical imaging. It is believed that the techniques of distribution of function and sharing of resources, such as computer power, display stations, and long-term archiving, will prove to be as powerful for the other imaging modalities as has already been demonstrated within nuclear medicine. This paper details the history of the development, the current results, and future plans.

A Nuclear Medicine Department has implemented a central computer system which stores all gamma camera images on hard disc. Multiple terminals allow simultaneous acquisition, processing and archiving of these images from various narts of the hospital. Downtime has been negligible and a month's work of patient studies is easily stored on line.

Several PACS prototypes are currently being constructed across the country. This article traces through the development of one such prototype at the University of Kansas. The prototype consists of three types of nodes; an image acquisition node, an archival node and a display node. These nodes communicate with one another via an Ethernet protocol. The reader will gain a view of the prototypes hardware and software architecture as it is to date. Motivations for key engineering decisions will be presented. Finally, the current status and future plans for the prototype will be discussed.

The National Library of Medicine (NLM) serves as both a national archive for medical literature and a national resource for current health science information. To fulfill this congressionally chartered mission NLM receives over 18 million pages of new literature annually into its 3 million item collection of print and nonprint material. The size of this continually increasing collection and its progressive deterioration due to the acid content of paper has encouraged NLM, along with others in the information field, to seek alternatives to conventional means for information storage. The Lister Hill National Center for Biomedical Communications, the research and development division of NLM, is developing an engineering prototype of an Electronic Document Storage and Retrieval system. Presently the system stores material on magnetic disc storage media, employs image capture stations using CCD array scanners, and image displays in both hard and soft copy high resolution forms. The next stages in design activities are being pursued to expand the system to include a mass storage optical disc-based archival system. This paper examines the design goals and organizational objectives leading to the development of the engineering prototype system.

At the First International Conference on PACS for medical applications, the fiber optic communication network in our department was introduced. Since that time, the network has expanded substantially, the communication protocols have been significantly modified, new fiber optic interfaces have been designed, built, and installed and even further hardware changes are planned. As we gained experience during this evolutionary period, we have acquired even more confidence in the use of fiber optics, the star network configuration, standard television for review and comparison, and the hierarchical communication protocols described in this paper.

Two fundamental problems face any hospital or radiology department that is thinking about installing a Picture Archiving and Communications System (PACS). First, though the need for PACS already exists, much of the relevant technology is just beginning to be developed. Second, the requirements of each hospital are different, so that any attempts to market a single PACS design for use in large numbers of hospitals are likely to meet with the same problems as were experienced with general-purpose Hospital Information Systems. This paper outlines some of the decision processes involved in arriving at specifications for each module of a PACS and indicates design principles which should be followed in order to meet individual hospital requirements, while avoiding the danger of short-term systems obsolescence.

The present paper is concerned in the general medical image diagnosis system that is now being developed by us. This system is called "DR SMIS" (abbreviation of "Digital Radiology by Sakura Medical Imaging System"). The system is to comprise five subsystems, namely, the Manager or image data base taking central role, and four others, Image Processor, Image Input Port, Image Hard Copier and Network. Of these five subsystems, the Manager, the image processing language for the Image Processor, the film digitizer for the Image Input Port, and the film laser writer for the Image Hard Copier have been completed till now. Here we will describe primarily about such image processing language system.

With the introduction of new radiological imaging modalities, especially nuclear magnetic resonance (NMR) scanners, an urgent need exists for an image communications/database management system which addresses not only intra- but also inter-departmental needs. Because space and shielding requirements will isolate NMR from the sites of other imaging modalities, a communications system is required to transmit and receive images from other radiological divisions and even clinical areas outside the department. Secondly, the variety and information content of NMR images obtainable from a single patient study demands rapid and easy consultation with a range of radiological and clinical specialists. Thirdly, patient scheduling and transportation require an integrated database management system coupled to a communications link so that optimal patient study sequences involving multiple imaging modalities can be routinely obtained. To meet these needs, a radiological information/communications system, called ICDBM , is being evaluated in a 1400 bed teaching hospital. This broadband video system integrates four CT scanners, two DR's, multiple scintillation cameras and a new NMR suite. ICDBM carries patient and departmental information as well as video and digital images (NTSC and up to 1000 lines). The rationale for system development, installation problems, and experience with ICDBM will be reported in detail. This paper not available at the time of printing of the Proceedings.

Medical imaging today is generally organized by anatomic systems (chest, neuroradiology, genito-urinary) or by physical technique of examination (CT, ultrasound, nuclear medicine). Interpretation of these examinations often requires comparison to previous studies utilizing the same modality or comparison to a study of the same organ using another modality; the concept of PACS includes the ability to ease the performance of these comparisons for digital image data.

This paper describes MODUS, an intelligent data management subsystem currently offered by NCR. This paper also discusses MODUS in an Image Database Management Subsystem. MODUS consists of the hardware and software necessary to communicate via a high level interface to hosts or host networks for the purpose of managing data and peripherals for those hosts. The hardware configuration is an Input/Output Processor, one or more Communication Processors and at least one SCSI Interface to peripherals. The software configuration consists of data and peripheral management Controlware, and multiple optionally supported communication protocols. MODUS will accept high level commands from hosts to access, store and maintain data. Data management features such as multi-user file and database access manage-ment, spacial management, back-up procedures, archival procedures, data security and data integrity are supported. Resource management features such as shared peripherals, hierarchy of memories including cache, standard SCSI interface to peripherals, error logging, print spooling and multiple peripheral types including memory, disks, tapes and optical products are supported. Multiple communication paths can be used with MODUS. Physical connections like NCR OMNINET, RS232 and Ethernet are supported. Communication protocols like CP/NET and Ethernet are available. MODUS would be used in an Image Database Management system for distributed image data storage and retrieval, archival storage management, image data availability and integrity and personnel and patient database management.

Digital storage and retrival of diagnostic image information in a large medical center is dependent upon the interconnection and successful use of many types of electronic equipment. Much of this equipment may be sufficiently dissimilar to cause major interconnection problems. For these individual systems to function together in a picture archiving and communication system (PACS) network it is essential that of their use and operation be coordinated in a systematic manner according to standardized procedures. There are several benefits to the adoption and use of such standards: a) New equipment and software can be designed for parts of the system without the entire system becomming obsolete. b) Permanence of the archived image material will be unaffected by replacement of archive equipment. c) Communication of image data between medical centers will be simpler. This talk will outline desirable characteristics for standards designed to help meet these objectives. Aspects of archive system operation that are suitable for standardization will be discussed and contrasted with those most likely best left unspecified to encourage creative diversity.

Much work is currently being done in the area of national and international data communications and computer networking standards. One highly visible activity is being conducted by the International Organization for Standardization (ISO) and the American National Standards Institute (ANSI) to develop standards for Open Systems Interconnection (OSI). However, most Local Area Network standards work to date has dealt exclusively with the Physical and Data Link Layers (Levels 1 and 2) of the ISO architecture. Upper level standards must be developed to fully meet the applications needs of network users. These upper level standards and protocols can provide such services as character echoing, error control, source/destination addressing, and encryption. Full specification of these and other upper level protocols is essential to the development of heterogeneous multi-vendor document retrieval systems.

There has been much discussion of the hardware associated with local area networks (LAN), but little discussion of the practical impact such a system will have on the practice of radiology as we know it today. This paper will address the problems associated with conducting consultations in a networked environment. The first consideration is what messages must be formulated. Questions such as the following arise: can one or more physicians communicate with a digital image source at the time a study is being done to instruct the technologist to obtain another view or to change the window width and level before the film is made; how can the surgeon viewing an image consult with the radiologist to obtain information concerning tumor localization; once the image is on the network in a standard format, what might be the best way to distribute it to two or more different locations in such a manner that consulting can occur effectively; if two physicians are consulting how can a third person either be included in the conversation or prevented from listening in. This paper will focus on the messages which must be exchanged to communicate the necessary information and to control the sharing and cooperative use of this information.

We discuss the requirements of a Picture Archiving and Communication System (PACS) for medical, from the user level. Based on the current requirements and current inadequacies with film file rooms, and anticipated future requirements, we suggest a minimal set of operation classes that a working PACS must support. No specific command set is proposed; each class is discussed purely in terms of what functions it provides to the user.

The term Image Workstation refers to the image display and processing systems of a PACS which are located within a diagnostic imaging department. They shall give access to pictures and data according to the work routines, thus being the interface between a PACS and its users. The paper will describe the role of the diagnostic image workstation. Evidently critical but most important will be all functions concerning image handling and display because of the principal differences between a physical image ('hard image') and a displayed image ( 'soft image'). The paper will therefore be focussed on the evaluation of required image functions and their general technical solution.

The rapid transition of radiology from film to digital imaging leads to new approaches in storing, communicating, displaying, viewing and retrieving diagnostic images. The purpose of this paper is to describe a prototype multiple digital viewing station being designed and implemented in our department as a first major step in this direction. The development involves four phases.

In 1975, when Dunn Instruments introduced the first commercially available raster scan film recorder, the device was used primarily with compound B-mode ultrasound scanners and produced 9 images on 8" x 10" photosensitive film. A charge-sensitive (Silicon 14) scan converter was used in the ultrasound unit to generate the raster display output to the film recorder.

Compression of digitized radiology images in a Picture Archive and Communication System (PACS) environment can reduce archival storage requirements as well as transmission rate constraints. Often data compression systems are designed for a communication system environment where the encoder must operate in real-time and have a limited complexity while the decoder may be more complex and operate off-line. A PACS network typically has different requirements. We present an overview of the current image compression literature relevant to a PACS environment. We consider both noiseless source-coding which requires perfect reproduction of the image, and source-coding subject to a fidelity criterion which allows some distortion in the reconstructed image. The importance of time-complexity versus compression tradeoffs is discussed.

Large digital files are inherent to CT image data. CT installations that routinely archive patient data are penalized computer time, technologist time, tape purchase, and file space. This paper introduces compression techniques that reduce the amount of tape needed to store image data and the amount of computer time to do so. The benefits delivered by this technique have also been applied to online disk systems. Typical reductions of 40% to 50% of original file space is reported.

Three dimensional arrays of data representing measures of human body tissue properties are produced with x-ray computed tomography, nuclear medicine, ultrasound and nuclear magnetic resonance imaging instruments. Array sizes vary from (64,64,64) to (512,512,128). Techniques to review the array values on a display screen include oblique plane, reprojection with selected dissolution, and simulated surface illumination display. The number of computer instructions required to generate these displays varies from 3.5 to 2500 million .The implementation of these methods requires large, fast random access memory (16 megabytes) and computers capable of executing a minimum of 10 million instructions per second. While computationally expensive, the use of three dimensional display techniques can be essential for accurate disease diagnosis and for optimizing disease treatment.

The use and implementation of digitally formatted imaging instrumentation for imaging disease is increasing. The amount of digital images to be displayed is huge, growing at an annual rate in Radiology Departments at eight percent. The use of local area networks for the acquisition, display and manipulation, and archiving of these digitally formatted images requires the careful design and implementation of Image Display and Manipulation Nodes. This paper addresses the amount of image data to be displayed and the required interactive graphic functions.

Although the new imaging modalities in radiology frequently use the terms "image" and "information," their meaning for the radiologist can be radically different from that for other scientists. Phenomenology (and hermeneutics) as an applied philosophical method can offer a much-needed clarification of these terms by examining the professional context of their use. Each profession accepts a different type of evidence for its professional judgments, evidence that is largely dependent on its specialized instrumentation. The paper will give examples of reliance on evidential meaning and illustrate the hermeneutic process of reading signals for global interpretation. While different professions will utilize different signals for their interpretations, the selection process involved is far from arbitrary; in each case it is a valid method of integrating individual patterns of information into a coherent whole. Phenomenological procedures can provide assistance toward reducing misunderstanding and frustration arising from the differences between the clinical world perception of the radiologist, on the one hand, and the scientific one of the physicist, on the other.

This paper describes a device capable of image processing, including subtraction using LCLV's and discusses its feasibilty as a clinical image processor. Intensity subtraction may be accomplished by the superimposition of two image; one a negative, and one a positive. The negative and positive images necessary for subtraction are produced by the LCLVs . It is also possible to perform " color subtraction " in which the difference image is represented as a pseudo-color distribution. Other image " operators " and their applications are discussed.

The parameter, quality of radiographic film exposure, was evaluated by computer and 15 human observers for the purpose of determining the feasibility of automatic computer eval-uation of radiographs prior to archival storage. Histograms of digitized knee radiographs were parameterized according to the Pearson technique and the computer was used to grade exposure quality relative to a set of training radiographs evaluated by the readers. There was a significant variation in the assessment of film exposure quality among the different observers. The computer scored 73% for the AP films and 53% for the laterals relative to the general consensus. The discrepancies were explained in terms of differences in anatomy and radiographic technique. The significance of this work is that it demonstrates that a computer can grade films according to a predetermined standard set by the radiologist.

This paper discusses our implementation plan for a digital image processing and viewing network in a large radiology department. We have selected a broadband coaxial cable network as the transport medium with a partial decentralization of computing facilities. We will also be implementing the system in the department in a series of phases designed to provide a smooth transition from present methods to network utilizing ones. Our reasons for choice of transport medium and a rationale for the phases and their selection will be covered. Potential problems and contingency plans will also be discussed.

During the past four years, a large number of radiographic images have been interpreted in both film and video modes in an effort to determine the utility of digital/analogue systems in general practice. With the cooperation of the Department of Defense, the MITRE Corporation, and several university-based radiology departments, the Public Health Service has participated in laboratory experiments and a teleradiology field trial to meet this objective. During the field trial, 30 radiologists participated in the interpretation of more than 4,000 diagnostic x-ray examinations that were performed at distant clinics, digitized, and transmitted to a medical center for interpretation on video monitors. As part of the evaluation, all of the participating radiologists and the attending physicians at the clinics were queried regarding the teleradiology system, particularly with respect to the diagnostic quality of the electronic images. The original films for each of the 4,000 examinations were read independently, and the findings and impressions from each mode were compared to identify discrepancies. In addition, a sample of 530 cases was reviewed and interpreted by a consensus panel to measure the accuracy of findings and impressions of both film and video readings. The sample has been retained in an automated archive for future study at the National Center of Devices and Radiological Health facilities in Rockville, Maryland. The studies include a comparison of diagnostic findings and impressions from 1024 x 1024 matrices with those obtained from the 512 x 512 format used in the field trial. The archive also provides a database for determining the effect of data compression techniques on diagnostic interpretations and establishing the utility of image processing algorithms. The paper will include an analysis of the final results of the field trial and preliminary findings from the ongoing studies using the archive of cases at the National Center for Devices and Radiological Health. This paper was not available at the time of printing of the Proceedings.

This paper describes a set of feasibility studies on converting the Department of Radiological Sciences, UCLA from a film based operation to a digital based operation. The studies address the following topics: the departmental facility, the operating procedure and cost, a cost effective analysis and a proposed digital based operating system. In addition, we examine three prototype projects being carried out in our department. These include: a digital archiving and communication system, a hybrid archiving and communication system, and a multiple image viewing system.

Digital fluoroscopic image acquisition, processing and storage for ventriculography, coronary artery and bypass graft imaging involves many practical considerations. Issues of sufficiency in terms of spatial, temporal, gray scale resolution and intravascular contrast sensitivity arise when 35mm cineradiographic film acquisition is replaced by digital fluoroscopy as the primary imaging modality. We have qualitatively and quantitatively evaluated and compared digital fluoroscopic and cineradiographic systems for cardiovascular image acquisition, storage and display in the cardiac catheterization laboratory. A digital fluoroscopic system (Siemens Digitron I) was evaluated and compared favorably to 35mm cineradiography in temporal resolution. Spatial resolution of the digital system was poorer, but contrast sensitivity (with digital subtraction) far exceeded the capabilities of cineradiography. As further modifications are made to adapt digital systems to the cardiac catheterization laboratory, complete replacement of 35mm cineradiography by digital vascular imaging systems is likely in the near future.

Concepts of Data Base Management Systems (DBMS) have not yet been widely used for medical image data bases. At the Hamburg University Hospital these concepts are applied to design an image data base for a radiological department. A scenario is described in terms of conventional data base design techniques and thus a conceptual schema is derived. This schema is modelled by the relational data model. For the implementation of the model we used the relational DBMS ORACLE. The software structure has two layers: one which extends the data base language to the needs of image management and another one which provides interaction with the data base in a pictorial context. The properties of the designed system are discussed and the state of implementation is outlined.

A PACS database must manage three different types of data; structured data in the form of patient/exam identification information, and unstructured data in the form of text and images. Queries based on the content of text documents as well as the content of images must be suported, in addition to those based on standard, well structured keys such as name, age and sex. We model the PACS database as three logically distinct databases, each supporting one of these data types, with mapping structures relating all three. Several design issues which have a bearing, both on these models and on physical implemtations, are discussed. Because image database systems are the least understood at this time, most of the paper focusses on the them. We briefly discuss current trends in structured and text databases, without discussing commercial systems, and then present some current methods of implementing image database systems.

The volume of data produced by new imaging modalities has far outstripped the ability of most departments to effectively utilize the images produced. The problem is further exacerbated by the fact that the diagnostic procedures have become progressively less invasive and traumatic and are being applied to an ever larger patient population. The decrease in cost and the rise in technological capability of computer systems in recent years has provided imaging specialists with the opportunity to create network systems for the storage and recall of diagnostic images. This paper examines the philosophy of image storage from the standpoint of the medical, legal, and practical questions. A proposal is made that not all images are equal and that some deserve to be archived for longer periods than others. The practical problem of using a video display for diagnostic readout, aside from the classical questions of resolution and response time, is discussed. A proposal is also made that two data bases might be created; one which provides rapid access to the clinically relevant images (i.e., the two or three that demonstrate pathology) and one which may require much longer to access, but which contains all the archived data.