This session on Film-Screen Sens item etry will be concerned primarily with the image receptor most used in medical radiography: the screen-film-process combination. We shall not, however, be dealing with its optical image-forming properties--only the response of the system to changes in exposure. One may liken sensitometry to the d-c response of an amplifier; the subject of the next session--Film-Screen Optics--may be likened to that of the a-c response.

This paper describes a system which approaches the state-of-the-art in energy dispersive x-ray spectrometry. Special methods for obtaining x-ray spectra from machines operating at high current levels and other phenomena which could cause spectral distortion are discussed. Entrance and exit spectra from some typical diagnostic procedures are shown. It is demonstrated that the exit spectra can be closely approximated by replacing the phantom with aluminum filtration. Since the sensitometric performance of imaging systems is influenced by beam quality, this technique can provide x-ray beams similar to those used in diagnostic radiology. The usefulness of the spectrometer for accurately measuring kVp and determining unknown elements of an intensifying screen is also illustrated.

The American National Standard Method for the Sensitometry of Medical and Dental X-ray Films, PH2.9, provides standard methods for measuring the speed and average gradient of medical and dental x-ray films. The most recent revision of this standard was completed this year. While the response of a film can be measured by this standard method, there are no provisions for measuring the response of an intensifying screen or of a screen-film combination. Therefore, the sensitometric parameters of the film are of limited utility to users of screen-film combinations. In an attempt to correct this situation, a task force was established last year by Subcommittee PH2-31 of the American National Standards Institute (ANSI) to pursue the establishment of a new and separate standard providing a method for the direct measurement of the response of screen-film-processing combinations. This paper is a report on the status of the work of that task force.

Sensitometric testing standards are needed in the medical x-ray field to fill many different functions. The radiologist, the manufacturers of film, intensifying screens, processors or processing chemistry, as well as governmental agencies charged with procurement of x-ray supplies or with monitoring x-ray exposure to the populace, all need reliable sensitometric information to assist them in making intelligent decisions. The kinds of information required by these various users are not identical, however, and are dependent on the end use to which it will be put.

For the evaluation of the imaging capability of radiographic film or film-screen combinations, it is important to know the density increments resulting from known exposure increments or, in other words, to describe the sensitometric properties of a given film or film-screen combination. It is customary (Refs. 1,2) to describe a film or film-screen combination sensitometrically by means of the so-called characteristic or H & D curve: For the sake of convenience, density is usually ploLLed as a function of the logarithm of relative exposure.

The panelists along with the speakers today will try to field the questions that are presented; however, I believe that before that begins some of the panelists would like to make some comments on their own: Dr. Clark will begin:

The optical characteristics of screen-film, combinations have a fundamental influence on radiographic image. quality because they affect sharpness, resolution, and quantum mottle. They are also linked to system. speed. Thus, screen-film optics can determine. detail visibility in the radiograph and patient exposure.

The optical and noise properties of radiographic screen-film systems have often been evaluated by concepts and techniques related to Fourier analysis. Figure 1 illustrates the physical characteristics of these properties. The point spread function (PSF), which is the two dimensional representation of the optical property of the screen-film system, has not been applied experimentally for evaluation. The screen-film system is generally considered to be isoplanatic and isotropic; therefore, the line spread function (LSF) in an arbitrary direction can be used. This function is much simpler to measure than the PSF. In the spatial frequency domain, the optical property is evaluated by the modulation transfer function (MTF), which is the absolute value of the Fourier transform of the LSF. Other optical characteristics such as the edge response are usually related to the LSF or the MTF.

In this paper we present a method for describing the performance of radiographic systems. Such a description can be given in a number of ways. Radiologists describe the system as being more or less suitable for one or another type of examination, according to their own particular criteria, criteria which are, it must be admitted, not always well understood by the layman. In this connection it is logical to speak of a "diagnostic quality criterium".

The major goal of medical imaging systems should be the production of an image, which affords maximum diagnostic accuracy while staying within acceptable radiation dose levels as well as acceptable economic and practical constraints. The ultimate test of an imaging system is not its aesthetic appearance, nor necessarily the specification of a particular physical measurement, but rather its diagnostic efficacy within the clinical setting.

Well ladies and gentlemen, Dr. Rossmann, I would like to follow in the same line as presented this morning in the keynote address by Dr. Elder and in the opening remarks for this session presented so clearly by Dr. Rossmann. What I would like to address myself to is maximizing the utility of the radiographic test and minimizing patient exposure. Now, maximizing utility -- it has to be utility as to diagnostic information. What I would like to address myself to is an area of investigation that actually starts by considering the MTF as the starting point, and the end point is the recognition by the radiologist of some pathological or anatomical detail of relevance.

1. As a rule it is a difficult task, with any imaging device, to combine high quantum detection efficiency with a high degree of fidelity to the original image. This is particularly true of X-ray imaging systems, due to two main factors: the strong penetrative power of X-rays through matter, and the complex of particle conversion processes involved.

For almost two decades x-ray image intensifiers have been widely--and now are almost universally--used for fluoroscopy and cine-radiography; increasingly, small format (70-105mm) static images, produced by photographing the output phosphor, have been obtained in those circumstances that in earlier days would have been accom-plished by larger format "spot films" made to document the findings of a fluoroscopic examination.

Modulation Transfer Function (MTF) is one of several key parameters that determine the overall performance capability of an x-ray image intensifier. The measurement of MTF described is made by illuminating the intensifier with a narrow line (Sifunction) of x-radiation and recording the line spread function of the resultant output illumination. The Fourier transform of this line spread is calculated to obtain the MTF. Measured MTF's are given from some typical intensifiers, on and off axis, in both the radial and tangen-tial directions.

A series of tests is described which can be used to determine whether image intensifier systems are operating properly. These methods minimize the amount of dis-mantling so that systems may be evaluated with minimum interruption of their regular use. New systems may be specified in terms of these tests to assure that minimum standards of performance are met at the time of system installation and acceptance. Suggestions are also given which describe the operating features of system components.

The quality of an image is evaluated by an observer by a subjective comparison with image impressions stored and remembered more or less distinctly. A photographic print for example, is judged by its gray-scale, sharpness or definition; and its graininess. The subjective ranking of "poor, good or excellent" depends on the size of the print and the viewing distance, and may change substantially when an excellent print is available for a direct comparison. A study of images on an objective basis has but one purpose: to determine objective criteria which will correlate with visual observations. It is evident that the characteristics of the visual system must be included in an objective evaluation of image quality.

At this stage of the game it is very difficult to add anything constructive because so much has been said; nevertheless I prepared some extemporaneous remarks about two weeks ago and they include seven points that I would like to make as quickly as I can. First, and I make these against the background of rather active participation in the TEPRSSC Committee, which as you'know is an advisory group to BRH as well as against the background of some 37 years in the x-ray industry. In the first place I sincerely and in a very heartfelt way commend the BRH for calling this session and commend the Society for the organization. There have been many times in the past and there will be many times in the future when the specter of Government regulations alarms me; however, there are also many times when the cooperation between Government and Industry can be extremely fruitful, and this is one of them.

Determinants of performance: There are four major factors which affect the imaging performance of a focal spot. I. Size. Intensity distribution. III . Shape. IV. Off-focus irradiation.

For high magnification radiography very small focal spot tube should be used. In 1953 we reconstructed the usual fixed anode tube (Toshiba Sealex, 10 kW) so that the potential of the focusing cup became lower than that of the heating fil ent by inserting resistence of 70 ka into the circuit containing the focusing cup and the heating filament. The electron beam e itted from the heating filament was biased and the focal spot of 0.15 mm in size was obtained. The elbow joint or the lung of the adult was roentgenograph ed in 2 or 3 times magnification(1). This type of X-ray tube has been developed step by step and at present 0.1 or 0.05 mm grid biased focal spot tube is available in Japan (Table 1).

One of the important causes of loss of resolution in a radiographic system is the finite size and shape of the x-ray focal spot (1). A knowledge of the limitations imposed by the focal spot is particularly important when one attempts to improve upon resolution by direct magnification (2,3).

In attempting to specify the performance of an x-ray tube focal spot, a number of factors must be taken into consideration which influence the spatial and intensity distribution of the focal spot: Tube potential, tube current, off-focus radiation, and off-axis variations. In addition, a knowledge of the tube loading capabilities is necessary to adequately predict the focal spot's role in the total imaging process. This paper presents techniques that were developed to investigate some of these factors and some typical results of work to date.

The focal spots of x-ray tubes used for diagnostic radiology are currently receiv-ing a great deal of attention. While researching the history of focal spot specifications, I was surprised to find the following statement under a dateline of September 17, 1 942 (Ref. 1 ): "The subject of focal spots seems to be receiving an undue amount of discussion these days. Some of the discussion is not based on fact and we think some fundamental principles might be helpful. "

It's been a very stimulating afternoon and there are many comments that I would like to make but I'm sure I won't have time for all of them. Our interest in small focal spots started from our interest in getting better images for angiography and for radiology it. general. We started out with the first magnification tube capable of serial filming made in this country by Machlett back in 1966 and from that point recognized in the clinical setting the great value that magnification played in many different areas of clinical application. We constantly pushed for improvements in development in this area and this led to the development by Bill Holland of Machlett of the technique of biasing focal spots so that one could manipulate the distribution from standard line type focus double peaking distributions that we were used to looking at and that Dr. Milne and Dr. Doi brought to the attention of the radiology world was not an ideal way to image. With the use of the very versatile application of biasing we then looked toward what was the practical question and that was what type of outputs could we get from these small focal spots--because in the real world the factor that we have to deal with is motion, a fact that has only been alluded to in this conference very briefly, and I must say that the Japanese patients must be smaller than the American patients because with the focal spots of that size we could not begin to get stationary images at exposure times greater than 80 milliseconds.

When selecting an X-ray tube, two factors are important for the radiographs sharpness "the tube power, responsible for a short exposure time and thus the sharpness when representing a moving object; and the geometrical image blurring due to a less than ideal focal spot."

The results of some preliminary measurements of the modulation transfer functions of (1) micro-focal spots of prototype x-ray tubes and (2) rare-earth oxysulfide intensifying screens are presented. The micro-focus tubes when used in conjunction with the rare earth screen-film systems have sufficient output to stop biologic motion in most clinical examinations. Optimum magnification is arrived at by use of the "effective sampling aperture" as suggested by Wagner (Ref. 1). An optimized micro-focus-rare-earth system is then compared to a conventional system of similar speed using calcium tungstate screens and 0.3 mm focal spot. The results show that a considerable improvement in contrast transfer is obtainable with the new system. Supportative radiolpgical evidence is presented.

In order to systematically improve a diag nostic system it is desirable to meaningfully specify the objectives. In dental radiogra-phy it is practically impossible to obtain absolute agreement as to what these objectives are or should be. At least two, however, are obvious and have essential limits which provide a sound basis for discussion. They are respectively: 1) The reduction of hazards associated with the use of ionizing radiation, and 2) the improvement of diagnostic accuracy.

Many popular examples exist for demonstrating how the statistics begin to show (and picture quality deteriorates) when atempts are made to form images with smaller and smaller numbers of photons (Ref. 1, 2). That this state of affairs reigns even in general purpose screen-film radiography has been shown over a dozen years ago by Cleare et al. (Ref. 3) and by Rossmann (Ref. 4, 5). The expression "quantum mottle" was invented to describe the blotchiness or clustering of film grains through which this effect is manifested. Quantum mottle, together with two additional but lower level sources of noise, namely screen structure mottle and film graininess, have been quantitatively analyzed in a series of experiments by Doi (Ref. 6). A few more examples of quantification of such radiographic screen-film noise can be cited (Ref. 7-10), but the samples studied are few and generally unidentified.

Electrostatic images are being used with greater frequency in diagnostic radiology. The application of Xeroradiography to mammographyM is the first potentially widespread possibility. Both inefficient charge recovery and the relatively poor quantum efficiency of the solid state absorber employed in the Xeroradiographic system have limited its use in general radiography, although improvements appear possible(2, 3).

Xeroradiography is an x-ray modality with image attributes distinctly different from film. These attributes result from the electrical nature of the process in contrast to the chemical nature of the film process. More specifically, they result from the in-herent relationship between the latent image and the electric field resulting from the latent image. The field acts upon the developer in such a way that image elements representing sharp structures and edges are enhanced and broad area elements are subdued. This aids in the visualization of such breast architectural features as the skin line, masses, vascular structure, and micro-calcifications. It also aids in the visualization of fine bone fractures, detecting foreign bodies in soft tissue and in the examination of casted extremities. In this paper some of the attributes of xeroradiography which provide diagnostic useful information, the origin of edge enhancement of xeroradiographic images, sensitometric methods and some approaches to the measurement of sensitivity will be described.

I hope that we all agree that the goal of radiologic imaging is good diagnostic accuracy combined with low radiation exposure. I am dismayed, however, whenever I hear the statement that we are all working to "Maximize diagnostic accuracy and minimize patient exposure" because I feel strongly that this statement somewhat subtly clouds a central and crucial problem in the evaluation of diagnostic image quality. I believe that we can never hope to maximize diagnostic accuracy and minimize patient exposure simultaneously because we can always buy a little more diagnostic accuracy at the expense of a little more patient exposure, or reduce exposure a bit at the cost of some loss in diagnostic usefullness.