The problem of reconstructing three-dimensional objects from a set of two-dimensional projected images has arisen and been solved independently in fields ranging from medicine and electron microscopy to holographic interferometry. By using a source of radiation external to the object, we obtain a transmission picture or projection of the three-dimensional object onto a two-dimensional surface such as the film of an ordinary electron micrograph or x-ray. The reconstruction problem is: Given a subset of all possible projections of an object, estimate its internal density distribution.

3-Dimensional Structure Analysis or image reconstruction is reviewed in view points of physics. Optimum energy of probe radiation in particular reference to photons and related aspects such as image enhancement using reference materials (-water like) are discussed. Existing computer algorithms are briefly reviewed in particular reference to the linear superposition with compensation (LSC) technique.

In order to accurately calculate dose distributions for patients in radiation therapy, it is necessary to define and consider in the calculations any inhanogeneities present in the irradiated volume. This study evaluates the mathematical foundations of a new method to determine the local densities (attenuation coefficients or electron concentrations). This method is based on measurements of the way in which the body section transmits high energy photons for a number of angular projections and a mathematical transform to reconstruct the point-for-point density. Several such reconstruction methods have been presented in the literature. This work deals with the application of these techniques to the radiotherapy problem by investigating which transform is most suitable and what the requirements are on the measurement technique. Four algorithms for reconstructing the attenuation coefficients within the body have been compared: 1. the convolution method; 2. the use of a simple numerical solution of the Radon transform; 3. the use of the fast Fourier algorithm to implement the Fourier transform reconstruction method; 4. the iterative Algebraic Reconstruction Technique (ART). The comparison has been based on the precision of the reconstruction method when applied to computer simulated transmission data. The problem of statistical fluctuations in the transmitted signal has been considered as has the problem of scattered radiation reaching the detector. Patient dose has been evaluated and the number of transmission measurements needed for a particular resolution has been determined. The results show that the first three methods give reconstructions that are quite similar in terms of their faithfulness of reproducing a known distribution of attenuation coefficients. The ART method did not perform as well as the others. Among methods (1), (2) and (3), the fast Fourier transform technique proved to be some 15 times more efficient in terms of computation time. Our findings also indicate that, for the radiotherapy problem, a total of about 25 projections are sufficient for obtaining an acceptable resolution.

Due to circumstances beyond his con trol, Dr. Ledley could not be here. My name is William Kiker. I am Head, Radiological Physics Division, Armed Forces Radiobiology Research Institute, Bethesda, Maryland. I am also an informal consultant in radiation physics to Dr. Ledley's institute.

In the present paper I wish to outline the physical principles under-lying the use of the phenomenon of Compton scattering in tomography(1)(2) (3)(4)and the reasons why not long ago it seemed like such a promising technique. Then, I would like to review the achievements to date, the limitations we have found, and by way of hopeful prediction, the ways in which these limitations may be over-come.

In the present practice of radiation therapy, a treatment regimen for a patient usually is developed by examination of dose distributions computed for one or more treat-ment plans proposed for the patient. Rarely are the dose distributions corrected for perturbations introduced by inhomogeneities such as lung and bone between the beam entrance surfaces and the treated region. These perturbations are not accounted for, even though they may affect the tumor dose by up to 20-30 percent in certain situations, primarily because the exact location and extent of the inhomogeneities within the patient are unknown. Various techniques have been proposed for delineation of patient cross-sectional anatomy and identification of inhomogeneities: (1) Visualization of cross-sectional anatomical displays from orthogonal roentgenograms projected in a special viewing device. This approach is complicated and provides no quantitive information for treatment planning; (2) Display of crosssectional anatomy on roentgenograms obtained directly by transverse tomography. The cost of transverse tomographic equipment capable of flexible patient position-ing and the poor quality of transverse tomographic images are the major limitations of this technique; (3) Representation of cross-sectional anatomy by obtaining individual patient contours and superimposing standard anatomical displays available in pictoral atlases. This method is no more than a crude approximation, because it offers no compensation fort anatomical variations among patients; however, limitations in these techniques have prevented their adoption in more than a few radiation therapy centers.

For the past year or so we have been working on the problem of locating brain tumors with ordinary hospital equipment and without the introduction of contrast material. Eighteen radiographs are taken at 10° angles around the head, they are read by a densitometer, and the numerical method of Kacmarz (1) is used to produce horizontal cross sections at any desired level. In principle this is also the method used by the EMI scanner, but of course the use of ordinary hospital equipment changes the practical character of the problem considerably. So far we have completed three experiments, two of which will be described in detail along with some of the problems that were en-countered.

Any object which is not opaque to x-ray radiation may be serially sectioned in any specified orientation (for any specified section thicknesses) without destruction,by means of internal density maps derived mathematically from the data provided by a series of ordinary radiographic images. The object must be rotated through 1800 about a given axis and radiographed at equiangular increments. Digitized intensity data obtain ed from the radiographs permit numerical analysis on a computer. The internal serial-section maps are derived by means of a suitable mathematical algorithm, and then drawn on a computer-controlled oscilloscope and photographed on microfilm. The numerical values for one set of serial-section maps (in our case, corresponding to sections perpendicular to the rotation axis) can also be stored on computer tape or disk to be later used to draw internal density maps for other desired object orientations. A single radiographic series provides all the necessary information to produce serial-section maps for any and all orientations of the object.

Ultrasonovision is a system for the measurement and visualization of ultrasonic fields which operates by interferometrically measuring the displacement amplitude of the acoustic wave. Its unique features include: high sensitivity, large dynamic range, wide angular response, good resolution, broad frequency range, and the capability for reflective or transmissive imaging. We shall describe the operation of the system and several areas of application, including the imaging and measurement of biological tissue.

Drs. Yokoi and Ito at Osaka University have developed several new methods for displaying more effectively a greater proportion of the received echo information than is currently displayed during B scan ultrasonic examinations.' Based on a visit to Japan and constant communication with Drs. Yokoi and Ito, this presentation will attempt to describe their work and suggest how it may improve ultrasonic diagnostic capability.

The technique of ultrasonic pulse-echo visualisation for medical diagnosis is well known and has received widespread acceptance in clinical practice in many diagnostic areas. The main areas of application at present are the abdominal organs (Ref.l,2), pregnant uterus (Ref.3) and heart (Ref.4). Clinical evaluation studies are being carried out in the eye (Ref.5,6), thyroid (Ref.7), brain (Ref.8) and breast (Ref.9). As the instrumentation and resultant echogram quality have been improved, the range of regions and conditions in which the technique has found application have been greatly increased. It is the aim of this paper to review the assumptions and physical laws upon which the technique is based and to examine the relationship between the limitations set by these considerations and the performance obtained with present equipment.

Edler and Hertz, in 1954, reported the first application of pulsed echo ultrasound to cardiac examinations. Since then, many studies have indicated the clinical usefulness of this technique establishing echocardiography as a valuable non-invasive procedure for the assessment of many cardiac abnormalities. Current clinical methods of echocardiography revolve about the utilization of A and T-M display mode ultrasound systems. These systems, although permitting visualization of cardiac motion, provide only one-dimensional echo information about the structures under investigation so that spatial geometries must be reconstructed mentally from a series of one-dimensional sonic interrogations.

The possibility of visualizing dynamic activity within the human body with ultrasound has attracted the attention of many investigators. Various approaches have been employed to gather and display the acoustic data depicting the movement of body organs. Visualization of the heart with acoustic energy presents a formidable challenge. The motion of the heart and its valvular components requires resolving valve leaflets moving at velocities of up to 120 mm/second. The task for the transthoracic approach is complicated by the fact that the heart is surrounded by lung tissue which is, for all practical purposes, opaque to ultrasound, and bone, which reflects and refracts acoustic waves.

Currently with the increasing use of medical ultrasonics there is a need for accurate characterization of diagnostic instruments. There are at least five professional societies in the United States actively pursuing some facet of this problem. There are also numerous foreign societies at work and, of course, the world standards organization, the International Electro-technical Commission (IEC). The IEC is the parent world organization and as such national and society standards should comply with the IEC standard.

Ultrasound is being increasingly utlized as a diagnostic tool in medicine. At the pres-ent time, ultrasonic diagnostic applications is one of the most rapidly growing areas in the field of medical electronics. Its use promises to rival x rays in the near future in terms of the number of individuals exposed. Along with the increasing use of this modality comes the responsibility and need for the correct assessment of ultrasonic field parameters. The evaluation of equipment performance is important in two areas: first, in the assessment of the clinical performance of the diagnostic machine, and secondly, in evaluating the potential risk from ultrasonic exposure in relation to biological effects. The documentation of performance parameters of current diagnostic equipment is needed to provide data for epidemiological studies as well as laboratory investigations. In terms of obtaining information related to the effect that ultra-sound may have on tissue, parameters of importance include average power, average intensity? peak intensity, spatial intensity distribution, pulse repetition rate,pulse length, and ultrasonic frequency. In this area, where uncertainties exist with regard to potential biological effects, the prudent course of action is to maintain exposure levels as low as is practical consistent with obtaining the necessary diagnostic information.

Today's use of ultrasound could be divided in two major areas: the therapeutic and the diagnostic. The major difference between the two applications is the ultrasonic power level at which the equipment operates. In therapeutic applications the systems operate at ultrasonic power levels of up to several watts per square centimeters while the diagnostic equipment operates at power levels of well below 100 milliwatts per square centimeters. The therapeutic equipment is designed to agitate the tissue to the level where thermal heating occurs in the tissue and experimentally has been found to be quite successful in its effects for the treatment of muscular ailments such as lumbago. For diagnostic purposes on the other hand as long as sufficient amount of signal has returned for electronic processing, no additional energy is necessary. Therefore considerably lower ultrasonic power levels are being used.

There are several groups interested in the development of performance measurement criteria in the diagnostic ultrasonic Area .These include, first, researchers studying the biological effects of ultrasound, secondly, manufacturers who have the need to supply prospective purchasers with definitive information concerning equipment performance as it may be related to potential biological effects, and thirdly, groups who have a responsibility to evaluate available information and develop measurement methods and establish performance criteria.

It is generally accepted that diagnostic ultrasound as presently employed has no adverse effect on the patient, and that the margin between diagnostic levels and damage thresholds is large. However, the size of this margin is not known due in part to the difficulty of providing a high enough intensity to actually produce a measurable change in tissue with the pulse regimes used in practice. It may be useful to review the types of regime used in the various techniques and the threshold levels currently known.

With the exception of those involved in developing standard methods for measuring acoustic intensity, researchers studying biological effects of ultrasound have the greatest responsibility for accurate measurement of a variety of indicators of acoustic exposure. Classed among these researchers should be those involved in epidemiological studies designed to reveal any toxic effects of ultrasound at diagnostic intensities.

I'd like to make a comment, namely, that temperature can be an important dose parameter. Normally, temperature doesn't have to be considered in diagnostic procedures where the temperature of the patient is fairly uniform; but under abnormal circumstances, where the patient's temperature may be elevated or surpressed, it may become a consideration.

Image evaluation in diagnostic radiology and nuclear medicine is an area of concern to many researchers today. It is not only a challenging field for the physicist but of extreme importance to the radiologist. Ultimately both the radiologist and the physicist would like to optimize the radiographic imaging processes, but before it is possible to optimize a system it is first necessary to be able to quantitate the system in meaningful terms. An extensive effort has been directed to the quantitation of image quality in radiology over the past years utilizing objective measures such as the modulation transfer function (MTF), resolution, contrast, signal-to-noise ratio, etc. Since these do not include the interaction of the radiologist with the radiographic image, the results, more of-ten than not, do not correlate with the radiologist's subjective opinion of the system. More recently several researchers have applied the principles of psychophysical evaluation to radiographic imaging bringing the radiologist and his subjective interactions into the evaluation of image quality. This has led to some insight into the problems of radiological imaging, but the expense in terms of radiologist's and physicist's time may not be justifiable in many instances.

The ability of the eye to perceive detail in images is a complex function of image contrast, brightness and noise. Image amplifiers can remove the brightness limitation but noise sets the final limit on performance which can be obtained with weak signals. The manner in which weak signal noise limits the performance of imaging systems is reviewed. Image amplifier gains required to remove the masking effect of additive amplifier noise is calculated as function of image contrast. Finally, the limitations of finite system aperture response on overall performance is described. It is concluded that increased resolution capability and displays having greater dynamic range are needed to improve the ability to see fine image detail at low contrast in characteristically noisy x-ray images.

The problem of relating objective measurements of signals and noise encountered in diagnostic imaging procedures to empirical measures of observer performance, such as Receiver Operating Characteristic (ROC) curves (Ref. 1) is both important and difficult.

With only a handful of physicians in this large audience, I find myself in a rather pre-carious position trying to relate a complex topic to a group of scientist primarily from the physical sciences. Perhaps I can make an analogy. We spent a significant amount of time this morning elaborating on signal to noise ratio. If one looks at the spectrum of signal to noise ratio, in trying to quantitate factual information regarding the physical sciences on one hand and the biological sciences, particularly medicine, on the other, the results are rather interesting. There is no question in my mind, having spent a few years in physics, that the signal to noise ratio in the physical sciences is quite high. As one approaches the biological sciences, it diminishes significantly and as one approaches medicine it becomes quite small, particularly in fields such as psychiatry where information is extremely difficult to quantitate. The reason for this is relatively simple; in physical sciences an equation needs to be solved, an integral needs to be worked out or mathematical postulate is either proven or not proven. However, as one approaches biological systems, the number of variables, one depending on another, sometimes almost appears infinite. Therefore, to try to take some of these physical measurements, used in the day-to-day analysis of mathematical equations, etc., and use these tools and each of the variables in our biological systems sets up an extremely complex situation to solve. Therefore, for me to talk about the psychophysics from a physician-radiologist point of view is somewhat presumptuous on my part. However, our challenge in medicine today is to use the tools of the physical sciences to better quantitate these variables and it is for this reason that I appear before you today.

DuPont Cronex 4 film is approxi-mately half the speed of Kodak RPR film, and requires twice the number of photons to make an exposure. Consequently, quan-tum mottle should be less with Cronex 4 film, at least if the same screens are used with both films, as in our study. However, do you feel it is valid to compare a high speed screen system to another system which uses medium speed screens?

Starting with a brief history of the two stage diagnostic radiological technique, the development of x-ray image intensifiers with the aim of achieving improved fluoroscopy, with reduced exposure rates is discussed. The characteristics and limitations of these older image intensifiers are given. After these considerations about the performance of contemporary image intensification systems, the possibilities of improving these devices, as regards quantum detection efficiency, improved electron optics design using modern high speed computers, and improved output phosphor screen preparation are considered. In particular a discussion of sodium activated cesium iodide as the input scintillator, first reduced to practice by this author in 1967 is given. Some preliminary results on an experimental 210 mm input 70 mm output x-ray image intensifier tube are given which suggests that such a tube with a 100 mm output might be an adequate replacement for the presently used x-ray image intensifier screen-film combination.

This essay is intended to con-vey some reactions to the question "Image enhancement in radiological diagnosis - can it be effective?" submitted to a panel at the SPIE meeting in November 1973. The reac tions of persons who are directly involved in the mechanics of image processing are colored by optimism fuelled by acceptance or pessimism fuelled by neglect. This observer represents the radiological user and therefore may command the ob jective middle ground. Hopefully the future will be directed to types of enchancement for which greater rewards exist.

Since the discovery of I-131, thyroid uptake and scanning procodures have been a major radionuclide use in medicine and in part led to the development of the field of Nuclear Medicine. The first method of detection of the distribution of radioiodine in the thyroid was by simply moving a small collimated geiger tube by hand over the neck area and recording the count rate from radioiodine which had become located in the gland 24 hours after ingestion. This was an internationally used technique in the early 1950's. Since then, scanning techniques have developed in various ways and new radionuclides have been taken into use. The clinical demands On the result of scanning are essentially the same now as they were then. This is a review of pertinent aspects of scanning.

Ultrasonic waves are sonic vibrations of frequencies above 16 to 20 KHz which is the upper limit of audibility for the human ear. Above 15 MHz, ultrasonic waves are referred to as microwave ultrasound because at these high frequencies ultrasound tends to act like electromagnetic radiation even though particle vibration is still being produced. Due to practical limitations set by ultrasonic generators the upper limit of ultrasonic frequencies is around 500 MHz. The frequencies of biological interest are in the range of a few KHz to a few MHz.

In Anger camera imaging, probably the most important parameter of camera performance is field uniformity. For the majority of camera users, field uniformity is evaluated in a subjective manner by simply viewing polaroid field floods. Due to film lim-itations, viewing conditions and statistical limitations, it is questionable if count density differences less than 15-20% can really be identified. To shift from a qualitative to quantitative method of field uniformity determination, a photographic method using high count density 70mm images has been developed. For the majority of Nuclear Medical imaging procedures, the current instrument of choice is the Anger camera. Of the An camera performance parameters, probably the most important is field uniformity. For those few institutions with dedicated computers interfaced to an Anger camera, field uniformity can be assessed and even corrected by the computer. However, for the majority of camera users, no quantitative method of uniformity evaluation exists. To shift from a qualitative to quantitative method of field uniformity determination without an interfaced dedicated computer, a phogographic method using high count density 70mm images has been developed; similar to film isodore methods in radiation therapy.

Automated computer analysis of cine ventriculograms provides a rapid, reproducible means of analyzing visual data obtained at cardiac catheterization. At the present time several laboratories routinely do semiautomated volumetric analysis of cine ventriculograms. However, their method requires the cardiologist or a trained technician to manually trace the outline of the ventricle on each frame. (Ref. 1 and 2) This technique is time-consuming and prone to errors.

The continued high mortality rate (Ref. 1) from breast cancer has stimulated a large amount of research (Refs. 2,3,4,5) in recent years towards the development of systems for earlier diagnosis.

Electronic radiography is a recently developed technique in which the radiographic image on an intensifying screen or output phosphor screen of an image amplifier is intensified by a closed-circuit television system, then recorded on photographic film or on a magnetic storage medium. This technique has proven capable of recording images of quality adequate for a wide range of diagnostic examinations and procedures, but at a dose significantly lower than that of other techniques.(1) Recent advances in television technology and in radiogrqphic screen manufacture have led us to develop a high resolution electronic radiographic system for mammography. Presented in this article are measurements which have been carried out on components of the system to optimize the system for mammography.

I must assume that all of us here are fully aware that the prevailing practice with manufacturers of x-ray tubes is to mislabel their products. Similarly, I am sure that all or most of us here know that appallingly few radiologists have been aware of this practice of mislabeling of x-ray tubes with regard to focal spot size. For many years I was one of the vast majority of radiologists who believed what I read on the label. I accepted the label designation "1.0-2.0" or "0.3" as a mathe-matically accurate description of focal spot size of the x-ray tube to which the label was affixed. Some 4 years ago I discovered otherwise. For the past three years I have been probing the boundaries of the practical consequences of this mis-labeling. Important facets of mislabeling include the clinical side effects: The important need for informing the profession, and, inevitably, the process of redress necessary to attain accurate or real world labeling of x-ray tubes.

It is well known that the x-ray intensity distribution of the focal spot can significantly affect radio-logic images, for example, blood vessel images in angiography (Refs.1, 2). For an accurate evaluation of the effect of this distribution on vessel images, it is essential to use the line spread function (LSF), the point spread function (PSF), or the optical transfer function (OTF) of the focal spot. However, for many years it has been common practice to describe the focal spot by its size, which is looked upon as a single figure of merit for the x-ray intensity distribution of the focal spot. Two methods are relatively well known; one is to measure the pinhole image size by a subjective judgment (Ref.3), and the other is to use the first-zero spatial frequency of the OTF for the determination of an equivalent uniform LSF (Ref.4).

One of the important causes of loss of resolution in a radiographic imaging system is the finite size and shape of the x-ray focal spot (1-4). Other factors are patient motion and the limited resolution of the image receptor, primarily due to the lateral spread of light from the intensifying screens within the x-ray cassette. The relative importance of each of these factors is inter-dependent as shown below.

There are two methods to determine the dimensions of a focal spot: 1. Imaging with a pin-hole camera. 2. Measuring the geometric unsharpness.

The x-ray tube focal spot intensity distribution is a continuum of numbers describing the source strength as seen in some plane under some set of specified conditions. Focal spot "size" is one number which attempts to communicate something meaningful about the continuum of numbers, e.g., perceptible outside dimensions on a pin-hole radiograph of the focal spot, disappearance frequency for a specified magnification, size of uniform distribution giving the same disappearance frequency, etc. A user might like to know the "size" of a focal spot so as to rank it against others, and to predict its effect on an overall imaging procedure.

Would Dr. Rao comment on the magnitude of off-focus radiation as a function of tube type, and also on the effect of off-focus radiation on either the MTF or other description of system performance.

I appreciate the opportunity to chair this panel discussion and to introduce members of the panel. The topic for discussion is one of growing importance as automatic brightness control circuits are incorporated into an increasing number of image intensification fluoroscopic units. As most of you know, these circuits work by monitoring the brightness of the output phosphor of the intensifier or the current flowing from photocathode to anode in the intensifier, and using the resulting signal to adjust the tube voltage (kVp), tube current (mA), or tube voltage pulse width to maintain the output phosphor image at constant brightness. Automatic brightness control circuits are useful for routine fluoroscopy and essential if uniform densities are to be obtained from one frame to the next in cinefluorography. With the advent of 105 mm photospot filming as a replacement for conventional spot films, automatic brightness control circuits are being used in place of phototimers to obtain images of desired density.

Fluoroscopy and fluoroscopic procedures were performed within months after the discovery of x-rays in 1895. These procedures were conducted with a fluoroscopic screen which the examiner viewed directly in a darkened room under conditions of rod vision. When I entered the field of medical physics in the late 1940's, this approach to fluoroscopy was still being used, and it was a rather poor method of obtaining information about the patient. Nevertheless, a considerable amount of help in the diagnosis of disease was achieved with this conventional mode of fluoroscopy. In the early 1950's, image intensif-ied fluoroscopes became available, and immediately there was a spectacular increase in brightness and realizable resolution and a correspogding spectacular increase in information retrieval from the fluoroscopic process. The major factor in these advances was conversion to the use of cone vision rather than rod vision. Although the image intensifier provided a spectacular increase in information retrieval, there was not a startling decrease in patient dose, although there was some; perhaps a factor of two or three. Most of the signal amplification provided by the image intensifier was used to create a brighter image and improve information retrieval, and little was applied towards the reduction in patient dose. Perhaps never again in the medical use of x-rays will we obtain such a spectacular improve-ment in information retrieval with an accompanying reduction in patient dose. Of course, initially there was some opposition to image intensified fluoroscopy. One objection was to the small field; early intensifiers had only a 5 inch input screen and radiologists had been used to a big rectangular screen for viewing. Some radiologists had difficulty identifying where they were anatomically during the examination. Incidentally, this problem no longer seems to exist, as evidenced by the fact that the sale of 6 inch intensifiers far exceeds that of 9 inch intensifiers. At present there does not seem to be much interest in larger intensifiers. One factor that did cause concern with the advent of image intensifiers was the constant need to vary machine parameters such as kVp, mA, etc., to maintain an image of constant brightness with changes in patient thickness and density. This problem stimulated the development of image intensified fluoroscopy with automatic brightness control (ABC). This development occurred in the mid 60's.

The Automatic Brightness Stabilizer, ABS, is that part of the X-ray control system which keeps the light output of the image intensifier constant over variations of patient attenuation and system geometry. This permits consistent TV or film exposures during a series of fluoroscopic examinations. (Ref. 1) A properly designed ABS will operate with several constraints: 1. It will hold the image brightness constant for variations of patient thickness and attenuation. 2. It will ignore marginal flashes and the effects of normal coning. 3. It will preserve image contrast and minimize noise. 4. It will keep the operation within generator ratings. 5. It will effect a reasonable compro-mise between patient exposure and image quality. 6. It will keep patient exposure to below 5 R/min except when an override is used. 7. It will respond fast enough to track during an examination. 8. It will correct for system variables such as IA mode, camera rate, etc. 9. It can be shut off and held at a particular value (after reaching equilibrium) prior to injection of contrast media. 10. It can be shut off to allow manual control of factors.

As a prelude to the ABC Panel discussion I appreciate this opportunity to discuss the technical aspects and philosophy of one member of industry towards automatic brightness control of fluoroscopic image systems. I shall begin with a brief review of alternate approaches to ABC.

This afternoon, I want to describe to you some reactions from our clinical colleagues about how they feel towards automatic bright-ness systems. Perhaps the first point to be made, as has been pointed out, is that radiologists by and large do not understand what an automatic brightness system is or how it works. Some radiologists may have a "textbookish" recollection of what was taught to them at one time in physics, but they don't really have a working knowledge of automatic brightness control systems in general, or of the specific system they are using in particular. With the increasing trend towards pushbutton automatic x ray systems, radiologists are increasingly losing the opportunity to become knowledgable about the systems and about the effects of components of the systems upon image quality. These are concerns I have heard many radiologists express. They also express some concern for retaining manual control of fluoroscopy for certain examinations, and desire an automatic brightness controlled fluoroscopic system which also provides manual capability. Another prob-lem which has been mentioned during my discussions with radiologists is the problem of large area sensing. This problem has been discussed by several speakers already, and I have little to add to this discussion. There apparently is a problem concerning movement of the x ray beam over areas of the upper thorax, esophagus, and neck, and reception in the edge of the field of intense radiation which misses the neck. A third difficulty which was addressed by two of the preceding speakers is the relatively slow response time of automatic brightness controlled systems and the problems this creates when crossing interfaces.