Roentgen discovered x-rays in late 1895 and published his first paper on the subject December 28, 1895. Almost immediately it was recognized that blood and surrounding tissues provided similar x-ray attenuation and that studies of the cardiovascular system would require the use of some intravascular contrast medium. The first arteriogram was reported in January, 1896, by Haschek and Lindenthal who injected the arteries of a cadaver forearm with a medium that was largely calcium carbonate. It was not until 1927 that Moniz of Portugal reported the first successful arteriograms in patients. He studied the intracranial vasculature with iodinated contrast material which was injected by means of a needle placed in a surgically exposed carotid artery. In 1929, Dos Santos and his colleagues performed aortography with the contrast agent injected into the aorta by way of a translumbar approach. The same group also used one of the earliest pressure injections to ensure rapid delivery of the contrast material. These workers were using iodine as the atom which was incorporated into the contrast medium. Iodine has good biological compatibility and shows an abrupt increase in its x-ray attenuation coefficient at 33 kiloelectron volts and in the energy range commonly used in diagnostic x-ray images it is more efficient in producing x-ray shadows than lead. Iodine remains the contrast atom of choice.

Digital radiography, a term hardly recognized two years ago, has grown to become the talk of the radiology community and the excitement of many commercial companies. M2st of this attention has been directed toward digital subtraction intravenous angiography), although during this same time period, a variety of digital radiography apparatus and image processing techniques have been under development. In November of 1980 at the RSNA Conference in Chicago, three commercial digital angiography systems were announced by Philips, Technicare and ADAC Corporations. During this same time period, the University of Arizona was discussing the concept of a photo electronic radiology department², the University of Pittsburg and Stanford University were investigating line scan radiography^3,4 and approximately five laboratories were carrying out clinical IV angiography with digital video systems.^5-9 These developments followed basic research programs in digital electronic and computerized imaging at various locations around the world. ^10-18 In the spring of 1981 we attempted to review the state of digital radiography, focusing on the various detector systems and image acquisition approaches.¹⁹ Since that time, rapid advancements in digital radiography have occurred. A major conference was held on digital radiography at Stanford UniversityzO, a new area detector system for digital radiography was announced by Fuji Film Corporation, clinical testing began on the Picker line scan digital chest unit²¹, and improvements were made in selenium detectors for digital radiography. Several additional companies announced digital video angiography systems, bringing the total now to approximately 15 companies worldwide. Digital video subtraction angiography is now well established as an important clinical diagnostic procedure and a variety of improvements and extensions of digital angiography systems are now ongoing. Digital acquisition and storage systems are increasing in both speed and storage capacity. Electronic image storage, transmission and retrieval is being discussed in greater depth, as indicated by a recent conference on this subject in Newport Beach, California.²² These developments indicate that the potential for replacement of x-ray film techniques with digital, electronic systems is becoming increasingly viable both technically and economically.

In quantitative densitometry it is essential to understand the statistical and systematic errors which contribute to the difference between the true and observed densities of anatomic structures. To study these we have developed two types of low contrast phantoms: in one type the contrast with the surrounding medium is generated by electron density difference, in the other type the contrast arises from atomic number difference. The resulting densitometric errors in a digital fluorography unit arising from system nonlinearities and drifts, image field distortions, image field intensity nonuniformities, image noise, generated contrast, point response function, beam hardening and subject scattered radiation are presented and discussed.

An expression for the noise in digitized video images is derived assuming an arbitrary video data transformation. From this expression, the ratio of quantum noise to camera noise is detenained. Unlike the noise itself, this ratio is independent of the video data transformation. The other quantity independent of the video data transformation is the pixel contrast signal-to-noise ratio. Noise measurements for linear and logarithmic video amplification are presented and discussed. The use of noise measurements on linearly amplified video images for estimating the system signal-to-noise ratio and the noise equivalent detected photon flux is demonstrated.

Digital fluorography using temporal subtraction techniques has achieved excellent visualization of the vascular anatomy with intravenous injections of contrast medium. However, clinical studies have demonstrated that a major limitation to this method is patient motion. Movements such as swallowing, breathing, peristalsis, cardiac motion, and arterial pulsations cause artifacts in difference images which can reduce the diagnostic accuracy of the procedure. Many of these artifacts can be eliminated by the use of hybrid subtraction, a second-order technique which combines temporal and dual-energy subtraction. This procedure is described and compared to both temporal and dual energy subtraction methods in terms of data acquisition, iodine signal-to-noise ratios, dose, and x-ray tube loading. Images are shown to demonstrate the viability of this new technique.

A new digital radiographic image processing system has been developed. The important capabilities of this system include a large image format, high image quality, on-line operation for reading, processing, and writing, and a software package for various types of image processing. A non-linear filtering technique is introduced which improves contrast of radiographic image structure while maintaining low noise in low-density areas. Preliminary clinical results demonstrate the improvements in diagnostic image quality.

A vascular tracing technique is described using a digital storage method which optimizes the contrast produced from radiographic images using very small quantities of contrast medium. The effects of further post processing and the production of high quality hard copy in a multiformat camera are illustrated with a number of clinical examples. The implications of this technique are discussed, including the possibility of producing clinically acceptable digital angiographic images at relatively low cost. The effect of digital averaging on fluoroscopy dose reduction is also illustrated.

This paper describes experiments which explore the ways in which variations in film and processing influence noise through their effects on the sensitometric curve shape and silver granularity. These include comparisons of several film types and processing conditions for which total radiographic noise from screen exposures and granularity from light exposures are analyzed relative to the instantaneous gradient at the measured density. It is seen that film granularity may indeed contribute substantially to total noise under conditions of low quantum mottle visualization.

The limits imposed on the spatial resolution obtained with screen-film systems for low contrast images are determined primarily by x-ray quantum noise rather than by the spatial response (MTF curve) of the system. For 40 keV x rays incident on a HiPlus/XRP screen-film system, the x-ray quantum noise limits the spatial resolution in the density region of unity from approximately 0.2 to 1 line pair per mm for film density differences respectively in the region from .02 to 0.1. By comparison, the MTF effect over most of this region of density differences, produces less than a 10 percent degradation of the spatial resolution.

The application of intravenous angiography to arteriography is currently, and justifiably, receiving much attention ^1-5 The advantages of this technique, where applicable, over the arterial approach are clear. While the application of digital equipment and procedures to intravenous angiography has stimulated interest in this technique, it should be emphasized that subtraction intravenous angiography can be performed with film-screen techniques.^6,7 The results obtained using film-screen techniques are generally superior to those which can be obtained with currently available digital systems. Although film-screen techniques are relatively simple to perform, appropriate films and photographic procedures must be used for optimum results. Care must be taken in mask production to include all the relevant data from the base film. The subtraction image should be printed on a film with contrast characteristics appropriate to the study. We will describe necessary photographic considerations for doing intravenous angiography simply and routinely with film-screen systems.

With the present great interest in digital methods of subtraction for angiography, it is useful to compare the clinical quality of images produced by careful control of analog techniques using subtraction film. The advantages and disadvantages of digital electronic imaging are discussed and a comparison with film methods is made with particular emphasis on the spatial resolution capability of each system, and the clinical significance of the limitations of each of the systems. A spectrum of examinations is presented to illustrate the limitations of both methods.

Analytical expressions have been written for image quality in mammography. A multiparameter optimization has been conducted to find the mammography system requiring the lowest patient dose for a given image quality. The optimization is subject to constraints imposed by technology such as; power limits on tube focal spots, absorption efficiency related to detector resolution and others. The optimization permits system geometry, kVp, filtration, detector resolution, focal spot size and grid characteristics to vary simultaneously and self consistently subject to an exposure time constraint. Significant dose reductions compared to current mammography systems have been found without assuming radical technological advances.

One hundred and fifty conventional craniocaudad mammographic images representing various benign and malignant breast conditions were obtained and four separate copies of the original images were made, each having a specific and different change in resolution. Three experienced mammographers evaluated the randomly arranged original images and copies and recorded their analysis. The analysis included visibility of various anatomical breast structures and pathological lesions (-masses and calcifications), subjective image quality, and mammographic interpretation. The resulting 2,250 evaluations were analyzed by Receiver Operating Characteristics Analysis. No statistically significant difference in performance was demonstrated by the radiologists in the range of the resolution changes evaluated.

An evaluation of imaging systems with different resolutions was made using MTF and Contrast Detail observer performance studies. The different imaging characteristics were obtained by changing the collimator in a clinical gamma camera. Three collimators were used. The two evaluation tests (MTF and CD) agreed on the relative ranking of the image systems. We conclude that contrast detail tests are useful in spite of their shortcomings.

Fogging of radiographic films can be characterized by a photographically equivalent radiographic exposure dose (Ex-eq). It is shown that within the limits of the experiments Ex-eq is independent of the moment of fogging i.e. before or after the radiographic imaging, and of the radiation intensity levels in the radiographic beam in case of no-screenexposure, but is dependent in case of an exposure with fluorescent screens and fogging by light. A dependent Ex-eq causes apparent sensitometric anomalies. The study warns against darkroom light fogging mainly when unloading cassettes.

In digital fluorographic techniques the video camera must accommodate a wide dynamic range due to the large variation in the subject thickness within the field of view. Typically exposure factors and the optical aperture are selected such that the maximum video signal is obtained in the most transmissive region of the subject. Consequently, it has been shown that the signal-to-noise ratio is severely reduced in the dark regions. We have developed a prototype digital beam attenuator (DBA) which will alleviate this and some related problems in digital fluorography. The prototype DBA consists of a 6x6 array of pistons which are individually controlled. A membrane containing an attenuating solu-tion of (CeC1₃) in water and the piston matrix are placed between the x-ray tube and the subject. Under digital control the pistons are moved into the attenuating material in order to adjust the beam intensity over each of the 36 cells. The DBA control unit which digitizes the image during patient positioning will direct the pistons under hydraulic control to produce a uniform x-ray field exiting the subject. The pistons were designed to produce very little structural background in the image. In subtraction studies any structure would be cancelled. For non-subtraction studies such as cine-cardiology we are considering higher cell densities (eg. 64x64). Due to the narrow range of transmission provided by the DBA, in such studies ultra-high contrast films could be used to produce a high resolution quasi-subtraction display. Additional benefits of the DBA are: 1) reduced dose to the bright image areas when the dark areas are properly exposed. 2) improved scatter and glare to primary ratios, leading to improved contrast in the dark areas.

Digital intravenous angiography may replace a substantial fraction of conventional angiographic procedures in the near future. The major advantages of this technique are reduced patient risk, lower cost and the fact that it can be readily performed on an outpatient basis. As with all radiologic procedures, patient radiation exposure must be considered in the overall evaluation of risk versus benefit. To this end, the radiation dose to patients undergoing routine head and neck intravenous angiography have been measured.

A method has been developed for the analysis of left ventricular wall motion which obviates the operator's involvement in the relative positioning of the systolic and diastolic contours. This was achieved by maximizing the cross correlation function for the two silhouettes. The technique was compared with a standard method requiring the operator's definition of a long axis for the left ventricle. The results for 21 normal angiograms showed that with the cross correlation technique the confidence region of the wall motion curves was markedly narrower and the symmetry in the contractile pattern between the anterior and posterior wall segments was better than with the standard technique.

The generation of contrast signals through digital mask-mode subtraction is studied with an emphasis on the Low contrast situations. The effects of beam hardening, radiation scattering, and veiling glare are discussed. The linearity between the contrast signal and the iodine concentration is tested at contrast levels from 3.7 mg/cm³ to 370 mg/cm³ of iodine in a plastic tube of 2.7 urn in diameter. A xenon perfusion study of a canine brain is presented.

The performance of a digital radiography system is characterized by the spatial resolution limit, the pixel contrast signal-to-noise ratio, the contrast detail curves, and the rendition of contrast signal. The importance and measurement of each of these characteristics are discussed. An equation for contrast detail curves based on noise analysis and threshold detection theory is derived to predict and interpret the low contrast performance of a digital radiography system. A phantom designed and constructed for law contrast performance is described.

A low contrast phantom designed specifically for the evaluation of the image quality obtained from a digital fluorographic imaging device (DFID) is described. The design rational and the image quality evaluation procedure utilizing the phantom are discussed along with the test results.

The goal of tissue characterization via quantitative computed tomography can only be accomplished with careful control of certain parameters. A quality assurance program is described which includes daily measurement of certain essential elements e.g. Hounsfield conversion, linearity, noise and resolution. Secondary measurements of collimation, bed movement, dose, and sensitivity profiles are performed on a monthly basis. Daily measurements are accomplished in less than 15 minutes while monthly checks take less than 1.5 hr thus minimizing conflict with a busy clinical routine. Examples of rapid recording methods and measurement techniques are included.

An automated method has been developed to test the performance of the Somatom-2 CT-scanner. The approach is based on the utilization of the three-section phantom supplied with the device by Siemens. The phantom contains rectangular blocks, a water bath and aluminum strips. Using the blocks, the edge response function, the line spread function and the modulation transfer function are computed. From the water bath section uniformity and noise are assessed: the image is divided into 80 subdivisions organized into four bands, four quadrants and two hemispheres, and the mean and standard deviation are computed for each. The same water bath image is also used for the computation of the noise power spectrum. The computer programs are being converted to facilitate on-line applications.

A method for determining the Square Wave Response function (SWR) for Computerized Tomographic (CT) systems is described. A 3° segmented acrylic phantom was scanned with water and also air as a second material. The data for each row of pixels was fit to a sine wave by an iterative least square program. The amplitude, phase shift, frequency and "D.C." level of the sine wave were all allowed to vary. The SWR was determined for various modes of the GE 8800 CT/T scanner. Advantages and limitations of the method are discussed.

When the available CT projection data are incomplete, there exists a null space in the space of possible reconstructions about which the data provide no information. Deterministic CT reconstructions are impotent in regard to this null space. Furthermore, it is shown that consistency conditions based on projection moments do not provide the missing projections. When the projection data consist of a set of parallel projections that do not encompass a complete 180° rotation, the null space corresponds to a missing sector in the Fourier transform of the original 2-D function. The long-range streak artifacts created by the missing sector can be reduced by attenuating the Fourier transform of the reconstruction smoothly to zero at the sector boundary. It is shown that the Fourier transform of a reconstruction obtained under a maximum entropy constraint is nearly zero in the missing sector. Hence, maximum entropy does not overcome the basic lack of information. It is suggested that some portion of the null space might be filled in by use of a priori knowledge of the type of image expected.

A software system has been developed to support analysis of CT images for tissue densitometry. Analysis can be done on either whole organ areas in a slice open subregions of these areas. CT numbers can he calibrated to absolute density. The system has been used to analyze lung and liver density in animal studies and lung density in a limited number of patients.

Density resolution the accuracy of CT numbers) is generally recognized by radiologists w'ao interpret Children's, CT to be very poor. A CT scanning phantom was made. in order to document the brain attenuation inaccuracies which do occur and also to derive normal brain attenuation values for varying sized heads, given. the skull diameters and thicknesses. In scanning' this phantom, other factors, some of equal importance, to small head size, were found to affect the Hounsfield numbers of brain. The phantom was scanned in order to determine the magnitude of these specific factors, using the GE 8800 model scanner. After head size (412 to 25, H), the variables of the head support (up to 15 H) and centering within the field of view (6-23 H) were of similar importance, for small heads. Kilovoltage, software, and machine drift were less, important, although only kVp settings, of 105 and 120 were employed. Manufacturers may improve CT number accuracy if they recognize the relative, magnitude of the various factors which alter measured attenuation.

Sensitivity, specificity, and predictive accuracy have been shown to be useful measures of the clinical efficacy of diagnostic tests and can be used to predict the potential improvement in diagnostic certitude resulting from the introduction of a competing technology. This communication demonstrates how the informal use of clinical decision analysis may guide health planners in the allocation of resources, purchasing decisions, and implementation of high technology. For didactic purposes the focus is on a comparison between conventional planar radioscintigraphy (RS) and single photon transverse section emission conputed tomography (SPECT). For example, positive predictive accuracy (PPA) for brain RS in a specialist hospital with a 50% disease prevalance is about 95%. SPECT should increase this predicted accuracy to 96%. In a primary care hospital with only a 15% disease prevalance the PPA is only 77% and SPECT may increase this accuracy to about 79%. Similar calculations based on published data show that marginal improvements are expected with SPECT in the liver. It is concluded that: a) The decision to purchase a high technology imaging modality such as SPECT for clinical purposes should be analyzed on an individual organ system and institutional basis. High technology may be justified in specialist hospitals but not necessarily in primary care hospitals. This is more dependent on disease prevalance than procedure volume; b) It is questionable whether SPECT imaging will be competitive with standard RS procedures. Research should concentrate on the development of different medical applications.

We have been studying the performance of x-ray grids both experimentally and theoretically. Grid performance depends not only on its inherent design, but also on characteristics of the scatter field, air gaps from scatterer to grid and to the detector and photon energy. In this paper we have evaluated the theoretical performance of x-ray grids for a variety of scatter fields where edge effects due to the finite size of the scatterer are taken into account, as well as photon energies and air gaps. Scatter field definition is based on our own experimental measurements, as well as the Monte-Carlo calculations of Doi, et al. The results of this work permit a calculation from first principles of such well-known grid characterization as: Bucky factor, Selectivity and Contrast improvement factor. The dependence of these characterizations on the scatterer, air gaps, etc. can be determined from the calculation.

As first indicated by Otto Schade ¹ and now commonly acknowledged, the formation of an optical image results when a set of random events is sampled over a finite space for a finite time. Silver halide grains do the sampling in film image formation by absorbing light and "remembering" it for extended periods of time. With the human eye, sampling is done by the individual rods and cones on the retina and the memory or exposure time is on the order of a tenth of a second.

Linear tomographic technique allows enhanced visualization of objects in a single body section (focal plane) while blurring the other sections above and below. This is achieved by a coupled linear motion of the x-ray tube and film-bucky during exposure such that the central ray of the x-ray beam always passes through the pivot point (fulcrum) and the film centre. Under this type of exposure and geometry, the focal plane is parallel to the film-plane and the objects oriented perpendicular to the focal plane provide the maximum shadow relative to other orientations. A new microprocessor controlled linear tomographic device can offer conventional, tilted, stereographic, and panoramic exposure sweeps to optimize the focal plane object contrasts depending on the object orientation. The angular interval of x-ray exposures is selectable from the control panel. A functional description of the Linear Orientation Tomographic System is presented.

An important parameter in any image intensifier-based fluoroscopic or fluorographic imaging system is the percent contrast (also called the Veiling Glare) of the image intensifier tube itself. A photographic technique for the measurement of this important parameter in-the-field, developed by the author over the last six years, has been used over the past year to survey seventy-five (75) different image intensifier tubes. Results for the percent contrast measurements for this group of image intensifier tubes will be given. Based on these results, a rating system of "low", "low normal", "normal", and "high normal" is established. Different rating schemes are derived for image intensifier tubes installed before 1979 (average percent contrast = 70%) and installed 1979 to 1981 (average percent contrast = 77%). Pitfalls in the application of the photographic technique of image intensifier tube contrast determination, as well as crosschecks on the method, are also discussed.

The results of maintenance service contracts with different manufacturers of x-ray equipment for a period of one year are presented and analyzed. The data concerning the maintenance service are discussed for three types of such services: (1) Service contract covering only repair. (2) Service contract covering preventive maintenance and repair. (3) On call service.

Large area contrast loss in image intensifiers results in reduced performance at low spatial frequencies and may limit the imaging capabilities of such systems for structures of clinical interest having inherently low subject contrast. The contrast ratio of image intensifiers is a measure of large area contrast loss which may be evaluated under laboratory conditions by photometric techniques. Unfortunately such techniques are not readily extended to clinically installed systems. A technique for the measurement of contrast ratio on clinically installed systems, incorporating a film recording camera, is proposed which eliminates the requirement of having a precise knowledge of the film's characteristic curve. The results of measurements obtained with this technique on clinical systems, as well as comparison of this technique to photometric measurements under laboratory conditions are presented, along with a discussion of the associate problems and pitfalls.

The performance of a radiological system can be evaluated on the one hand by an objective determination of the quality of the produced image and, on the other hand, by the dose delivered to the patient. In order to measure these two factors in a single exposure a Kodak breast phantom has been modified so as to simulate the breast absorption. The dose distribution is measured with thermoluminescent detectors. By consideration of a theoretical model of the X-ray imaging in mammography, a single quality factor is computed from the contrast, the spatial resolution and the noise measured on the phantom image. We present results obtained in various working conditions, i.e. variable X-ray tube voltages, use of different screen-film combinations, use of a grid.

I'd like to discuss with you a method that we use in the image stability laboratory for prediction of the effect of residual thiosulfate on the long-term storage of films. The first half of this talk appeared in the April, 1982, issue of the Journal of Applied Photo-graphic Engineering. I'll just be showing selected examples of the data here. Anyone who would like to see all the numbers can find them all tabulated in the publication. At least for X-ray films, high thiosulfate causes D-min discoloration. The second part of my talk will deal with what this discoloration is, both chemically and in terms of its tendencies to appear even after current chemical analyses may indicate that it shouldn't happen. I'll be using stain buildup and D-min discoloration interchangeably. For our purposes, they will describe the same phenomenon.

NMR imaging offers the possibility of obtaining a wide variety of clinical data. As a result the presentation and specification of NMR images needs care, since small variations in the technique by which they are obtained can produce radical changes in their appearance. A number of images taken with similar, but not identical, procedures are used to illustrate this, as well as to indicate the scope of this technique.

Digital image data is now generated in Ultrasound, Nuclear Medicine, Computed Tomography, and Digital Fluoroscopy Imaging Instruments and is anticipated for nuclear magnetic resonance imaging and direct digital radiographic recording. The interconnection of imaging instruments for the common purpose of image display and analysis forms a diagnostic image analysis network. For current studies, a typical hospital network would have to handle 4 x 1011 bits of data per year. Systems used only for investigative purposes would have to handle 2 x 10 10 bits of data per year. The demand to display image data is estimated at 38.5 or 9 display requests per hour respectively for these two systems. These demands can be met by sharing the use of central display controllers through a video distribution system. The needs of these systems can be met by 12 or 4 display controllers respectively. In the future, diagnostic imaging networks are expected to be common place with inexpensive archiving on laser optical disks providing rapid access to a full year of examinations.

A key measure of any radiology department's efficiency in performing examinations is its percentage of repeated films. In reliably obtaining high-quality films, some percentage of repeats is certain to occur. Human error and equipment malfunction can never be eliminated. Poor patient positioning, patient movement, film artifacts, exposure problems, and processing problems are only some of the reasons for performing a certain view more than once. But keeping this repeat percentage to a minimum should enable the department to use less staff time and fewer films, and help ensure that patients wait a shorter time and receive less radiation exposure. In this way, reducing repeats should both lower the cost and improve the quality of patient care.

Information and images from multiple sources, including CT, Ultrasound, and Angiography are transmitted to a central processing center. The digital information has standardized formats allowing for data manipulation and image processing. Display is either in single or multi-image format and integration of patient information is by multiformat cameras. Initial experience using this system will be presented.

Digital video radiography (non-subtraction) has been investigated as a new technique for applications in general radiography. The imaging system consists of computer and array processor for 512 x 512 x 8 bit image acquisition and processing in combination with 14" and 9" image intensifiers and high signal-to-noise ratio TV cameras. A digital filter has been developed and implemented on the array processor which provides corrections for image intensifier-TV camera structure mottle and image shading. Preliminary clinical trials have been carried out for IVP and GI examinations. The wide dynamic range and high contrast sensitivity of the digital system allows windowing to view through dense areas in GI studies as well as the ability to visualize subtle contrast in IVP and GI examinations.

A new type of digital radiography system of very high contrast sensitivity and spatial resolution is described which is based on the use of six linear arrays of self-scanning diodes fiber-optically coupled to a phosphor screen. The high detail of the system results from the fact that 6144 discrete diodes, 1024 per array, scan a field of view of 6 inches wide. A contrast sensitivity five times greater than film is achieved due to the high dynamic range of the diodes combined with the scatter rejection associated with the slit geometry. The entrance radiation exposure per image is 100 mR but could be reduced well below that in the future. Initial clinical experience has demonstrated the advantage of being able to display a single image over a wide range of window levels and window widths at the same time having a high contrast sensitivity in both the dark and light areas of the image. The complete digital radiograph is taken in a second, however the motion unsharpness is held to a minimum by virtue of an effective exposure time of 8 milliseconds. Applications to digital chest radiography and digital intravenous subtraction angiography in over 30 patients have shown the clinical value of this new form of radiography.

Continuous Digital subtraction combined with intraarterial injections of contrast medium permits the display of arterial structures during real time fluoroscopy. This DSA "road map" facilitates selective catheterization and has proved useful in interventional procedures.

Until now the application of computerized digital imaging process has been concentrated in the area of imaging production. A digital imaging process technique has been developed in our experimental work. Based on our preliminary experimental results radiographs of a minimal exposure (about l/4 of the standard exposure) is digitized and can be regenerated with the digital imaging process reported here to yield a digital image with good resolution similar to the radiograph produced by 100% of the standard exposure. This digital imaging process can reduce exposure by a factor of 4 for a radiologic image and hopefully it could be developed for further reduction of radiation exposure per a given radiological study.

In this study the basic concepts of parametric imaging, in which a temporal sequence of images is replaced with a single static image based upon a selected parameter, has been modified and expanded. In the new method fixed zonal areas and bands are established within the imaged organ and the mean value for each zone is computed. The early analysis of the data from the human kidney shows the presence of a pattern that has high potential for the quantification of perfusion studies.

An increased use of digital image acquisition, storage, and display systems within Radiology has led to the consideration of integrated digital diagnostic imaging systems which can provide image data communication, archival, and analysis. The results of an information content assay for Ultrasound, Nuclear Medicine, Computed Tomography, and Digital Fluoroscopy examinations is reported in a manner which allows the generation of digital image data for integrated imaging systems to be predicted. The four predominant examinations for each modality and their relative volume are identified, the average number of images recorded per examination is reported, and the minimum number of binary bits required to store an image is determined taking into account the use of minimal array sizes and image compression algorithms.The resulting weighted average information content for Ultrasound,Nuclear Medicine, Computed Tomography,and Digital Fluoroscopy is 57.2, .341, 7.9,and 3.0 megabits per examination respectively.

A justification for the introduction of digital image archiving systems in diagnostic radiology departments is presented. These archiving systems will be an integral part of a digital Picture Archiving and Communication System (PACS). The performance of proposed digital picture archiving systems within an all-digital PACS environment has to be determined by how well the systems fulfill the clinical need for image storage and retrieval. Given today's technology, a distinction has to be made between short-, medium-, and long-term digital archiving needs.

Photoelectronic-digital radiology has become a means for image acquisition at the University of Arizona. Large amounts of high speed image data are generated routinely from rooms remotely located from the computer/display facility. As an alternative to coaxial cables, a realtime digital video fiber optic link was developed. This paper will discuss some design considerations. It points out advantages and disadvantages of fiber optics. Some system and network applications are discussed.

A logical consideration of the digital imaging process (6) is described using the characteristics of the film and screen of a radiograph and a non-linear compensation curve for each pixel between optical density value of 100% exposure radiograph and 25% exposure radiograph obtained experimentally and theoretically. A radiological image of less than 100% exposure can be generated using the point-by-point mapping to form a radiographic image which is similar to that of 100% exposure. The mathematic expression of the film to screen characteristics and conversion between the optical densities of the 100% exposed radiograph and that of 25% exposure are given. A possible potential application of the digital imaging process is discussed.

Not since the advent of x-ray transmission computed tomography a decade ago has an innovation in medical imaging generated as much interest as that currently directed to nuclear magnetic resonance. This technique, long a standby in chemistry and physics laboratories, promises to provide images of reasonable spatial resolution and exquisite contrast sensitivity. In addition, quantitative analysis of specific elements in selected regions of tissue may be possible. In developing a strategy for the acquisition of nuclear magnetic resonance, cost factors must be considered together with a realistic appraisal of a clinical facility as primarily a research or clinical unit.

The 1976 Medical Device Amendments to the Federal Food, Drug, and Cosmetic Act provide for the classification of a medical device intended for human use into one of three regulatory classes based on the extent of control necessary to ensure safety and effectiveness: Class I, General Controls; Class II, Performance Standards; Class III, Premarket Approval. Class III devices are those for which there is insufficient information available to ensure safety and effectiveness through General Controls and Performance Standards alone. New devices such as Nuclear Magnetic Resonance Imaging systems fall under Class III because they were developed after the date of the law's enactment (28 May 1976). Investigational studies involving human subjects undertaken to develop safety and effectiveness data for a post-enactment Class III device come under the Investigational Device Exemption (IDE) Regulation (21 CFR 812). This regulation distinguishes between investigations of devices that pose a significant risk to the human subject and those that do not. A significant risk investigation "presents a potential for serious risk to the health, safety, or welfare of a subject." Procedures for obtaining an IDE differ if the device does or does not pose a significant risk. The sponsor of a clinical trial, and ultimately the Institutional Review Board (IRB), have the primary responsibility to determine whether a certain clinical use of the investigational device represents a significant risk to the subject of the investigation. A finding of significant risk does not mean that a device is too hazardous for clinical studies, but it does mean that a formal application for an IDE must be made to and approved by the Food and Drug Administration (FDA) before a clinical trial can begin. If the device is deemed not to pose a significant risk, unless otherwise notified by FDA, the sponsor is not required to submit an IDE application to FDA. Instead, the sponsor and investigators must satisfy only certain abbreviated requirements including maintenance of certain records and reports. In addition, the sponsors must maintain IRB approval throughout the investigation, label the device in accordance with the IDE regulation, and ensure that the investigators obtain and document informed consent for each subject under their care.

Anatomical and functional aspects of the circulation that affect approaches to imaging blood flow are discussed. Previous approaches to NMR measurement of blood flow and NMR. imaging techniques that show blood flow dependence are reviewed. A new imaging technique that uses alternately selective and nonselective preliwinary inverting pulses is proposed in order to separate blood flow effects from relaxation time effects.

A major limitation to the use of NMR as a sensing technique in clinical imaging is the low intrinsic sensitivity. As a result, the spatial and contrast resolution and scan speed are limited, with design gains in one requiring trade-offs of the others. In recent years several fundamental sources of detection noise have been identified and evaluated for medical NMR imaging. They are: thermal noise in the receiver coils, loading by the body conductivity and capacitance, and deflection of radiation by conductivity gradients. In the present paper we review these sources of contrast degradation and several other sources, including tissue-generated noise and shot noise, and show how to refine the estimates for diffractive effects at the skin surface. A quantitative example of signal/ noise estimation is presented for whole-body imaging.

High resolution 31P nuclear magnetic resonance (NMR) spectroscopy has been applied to the direct, noninvasive examination of phosphorylated substrate metabolism in the myocardium of live rabbits. By the combination of field profiling gradients and a surface, or flat, NMR coil placed directly over the region of the thorax which contains the heart, spatially localized NMR measurements of metabolic function in live animals can be obtained. This technique, termed "topical magnetic resonance" or TMR, has been used to follow the effects of several physiological conditions on the tissue pH and levels of key, energy-rich phosphorylated compounds in the hearts of live, anesthetized rabbits. Changes in tissue content of adenosine triphosphate (ATP), creatine phosphate (CP), and inorganic phosphate (Pi) and the NMR line widths of these species have been observed in animals given appropriate doses of adriamycin for a five day period. These preliminary data demonstrate the potential of spectroscopic NMR techniques in the evaluation of disease states in organs and tissues within the body and the ability to monitor both toxic and therapeutic effects of drugs.

The speed and accuracy with which NMR imaging methods can produce images in the presence of an inherently low signal-to-noise ratio is a major issue in the development of NMR imaging as a clinical imaging modality. In this regard, the Fourier and zeugmatographic techniques are among the most efficient in that they gather and interpret NMR signals from all portions of the imaged region simultaneously. Fourier NMR imaging has usually been viewed as a process of obtaining NMR spectra in the presence of constant field gradients followed by interpretation of the spectral distributions as projected spatial distributions. An equally valid perspective interprets the noisy raw NMR signals (FIDs) as corrupted partial spatial-frequency mappings of the imaged object. Insights afforded by this viewpoint are translatable straightforwardly into theoretical formulations which permit 1) quantitative comparison of performance characteristics of existing Fourier methods (including the zeugmatographic methods), and 2) derivation of new methods with optimal spatial-frequency sensitivity and accuracy characteristics, or with reduced minimum imaging times. A brief mathematical development of this spatial-frequency-domain interpretation of Fourier NMR imaging is followed by derivation of a proposed NMR imaging method with imaging performance capabilities superior to those of existing Fourier NMR imaging methods. Simulation results supporting these theoretical results are presented. Use of other optimality criteria to derive other new imaging methods is discussed.

Images of the head, torso, and limbs have been obtained using the nuclear magnetic reso-nance (NMR) of hydrogen. Images are presented and the imaging technique and apparatus are described. The mode of imaging, a spin warp 2-D Fourier transform technique with T1 discrimi-nation through partial saturation, is discussed and shown to be less demanding of field homogeneity than other techniques. The radio-frequency (RF) magnetic fields and pulsed field gradients are shown to be below the recommended limits for power deposition and induced electric fields.

Preliminary results from cranial NMR examinations of 180 patients and 40 volunteers are discussed. Three different pulse sequences have been used to produce images with varying dependence on proton density, T₁ and T₂. Repeated Free Induction Decay (RFID) images which largely reflect proton density are rather featureless and show limited changes in disease. Inversion-recovery (IR) images whose contrast largely depends on differences in T₁ show a high level of grey white matter contrast. In addition acute haemorrhage is associated with shortened values of T but many other conditions such as infarction, infection, demyelination, oedema and thalignancy are associated with increased levels of T₁. Spin-echo (SE) images whose contrast largely depends on differences in T₂ show very little grey white matter contrast but highlight pathological change in a variety of conditions against the bland background of the remaining brain. NMR has a number of important advantages over CT in imaging the brain and appears likely to assume an important role in neurological diagnosis.

Lesions of breast, thorax and abdomen have been evaluated by NMR techniques. More than 200 patients have been evaluated using this modality. T1 measurements, spin-lattice relaxation times, from the regions of interest have been obtained and compared with values from normal tissue of the same patient whenever possible. Visualization of lesions by NMR imaging has been achieved in breast, the chest and the abdomen. Cases evaluated include examples of both benign and malignant lesions. NMR results have been correlated with CT, nuclear medicine, conventional x-ray and ultrasound examinations, as well as histological evaluations.