The techniques employed in the Nottingham whole-body NMR imaging system are described. Attention is paid to the design philosophy of the instrument, describing in detail the construction of the apparatus. Results obtained from approximately one year's experience of head scanning are described.

Preliminary results have been obtained with two NMR imaging machines. High resolution images of anatomical cross sections can be produced with good soft tissue contrast and several tissue characteristics can be measured quantitatively. Functional studies are possible using paramagnetic materials as contrast agents, and blood flow effects can be readily demonstrated. NMR imaging therefore combines several features of other imaging methods in a single technique, which promises to make it a versatile tool for radiology.

While developing a human size NMR imager, a small aperture imaging system is being used to characterize NMR responses in intact rats. The basic response of our imaging technique to the T1 and T2 relaxation times is presented. Based on this response, T1, T2 and relaxation independent "hydrogen" images are computed. These computed mages of normal rats demonstrate that most of the contrast in NMR images is due to the relaxation times. Work on some possible hazards in NMR is summarized.

The ability of living cells and tissues to function properly is directly linked to basic biochemical events. The term metabolism is used to describe the integrated, concerted cellular biochemical processes that support cell and organ functions. Indeed, metabolism is the fundamental basis for understanding function and physiology in living systems. Therefore, any method or methods that could provide localized metabolic information for clinical diagnostic use would be of significant value.

We report on the results of a collaboration between two independent research projects at the Buffalo and Stony Brook campuses of the State University of New York. At Buffalo we have been developing a software system for the detection and display of surfaces of organs and organ systems from three-dimensional reconstructions. At Stony Brook we have been developing hardware and software for the three-dimensional reconstruction of objects using nuclear magnetic resonance zeugmatographic imaging. The merging of these two modalities gives us a truly powerful tool for three-dimensional visualization. This is demonstrated by frames from movies illustrating the external and internal three-dimensional structure of organs, such as the brain and the heart, produced by applying the Buffalo display programs to the Stony Brook nuclear magnetic resonance reconstructions.

A description of the nuclear magnetic resonance imaging system at Stony Brook with whole body capabilities based on a .1 Tesla air-core magnet with a 62 cm bore will be given. Important considerations for full three-dimensional (3D) imaging from projections include static field homogeneity, linear field gradient strength and uniformity, adequate trans-mitter and receiver capabilities and rapid data collection and processing. Preliminary results of our efforts to achieve a flexible system with potential clinical applications will be shown along with images of the head and breast from living human subjects. Since the 3D image has isotropic resolution, the image may be viewed from any desired direction.

Information transfer in a screen-film system occurs over a limited region of x-ray exposures because of the non-linear response of film to light emitted from the x-ray fluorescent screen. Based on screen light emission and film response data, the information transfer efficiency of a typical screen-film system is determined as a function of x-ray exposure for different x-ray energies. The method of determining this transfer efficiency is described, and the x-ray exposure regions over which information transfer can occur with this system is delineated for different x-ray energies.

Lack of perception at high densities on radiographs and the influence of viewing conditions on it are well known. This lack may be caused by blinding effects, by high visual noise at low light intensities or by a third phenomenon i.e. the dependence of the sensitometric properties of film on viewing conditions, which is analyzed in this paper. Reflection of ambient light by the film mainly lowers dramatically high densities and film contrast at these densities. Sensitometric curves of several films were measured under different viewing conditions by means of a telescopic photometer. The curves also can be deduced from curves measured by a regular densitometer when the optical properties of the film, the ambient light level and the light intensity of the negatoscope are known. The influence of the phenomenon under typical viewing conditions for the Curix MR4-film is demonstrated by means of sensitometric- and perceptibility-curves.

Inverse-square sensitometry for low kVp techniques is limited by air attenuation and beam hardening. It is thus difficult to obtain accurate H and D curves for mammographic screen-film combinations at technique factors used clinically. Methods are described for the determination of characteristic curves at low x-ray energies which provide composite intensity-scale curves using small changes in source-to-receptor distance and concomitant changes in either exposure time, beam attenuation or kVp. H and D curves obtained at low kVp's using the various sensitometric approaches are compared with those obtained using inverse-square sensitometry; the relative merits of the different methods are discussed.

Anyone who has tried to interchange different intensifying screens for a given procedure while maintaining consistent image density, appreciates the problem resulting from the lack of standardized screen speed designations. The problem has resulted, in part, from the variety of product names which we manufactur-ers have coined for our screens to establish screen brand identity and differentiation. These names, in most cases, give no indication of relative screen speed. Presently, there are no standard measuring procedures which the radiographer can use for routinely determining screen speed. The greater non-linear KVp response of the rare-earth phosphors, in fact, makes the use of single number designations deceptive. Different methods for designating screen speeds will be examined and a technique for transposing between different screens will be discussed.

To find the reasons for unnecessary exposure to patients and in assessing the performance of a radiographic system one needs to know the exposure to and the attenuation of the various components of that system. We used a water phantom to simulate patients of different thicknesses and for each, using a standard technique, measured the entrance, exit and cassette exposure to produce an optical density of 1.0 on the film of a common film-screen combination. The ratio of entrance to exit exposure is an indication of beam quality and the ratio of exit to cassette exposure is the reciprocal of the attenuation by the material between the patient and the cassette. These ratios for our laboratory system are used as standards against which beam quality and table-top, grid and phototimer attenuation are assessed. Measurement of the cassette exposure needed to produce a film of unit optical density when the different components of hospitals' film-screen-processor systems are interchanged with our laboratory system is a simple method of assessing, under clinically simulated conditions, the relative performance of hospitals' films, screens and processors. Results obtained in hospitals across Ontario will be presented.

The requirements for radiographic imaging phantoms for observer performance testing include realistic tasks which mimic at least some portion of the diagnostic examination presented in a setting which approximates clinically derived images. This study describes efforts to simulate chest and vascular diseases for evaluation of conventional and digital radiographic systems. Images of lung nodules, pulmonary infiltrates, as well as hilar and mediastinal masses are generated with a conventional chest phantom to make up chest disease test series. Vascular images are simulated by hollow tubes embedded in tissue density plastic with widening and narrowing added to mimic aneurysms and stenoses. Both sets of phantoms produce images which allow simultaneous determination of true positive and false positive rates as well as complete ROC curves.

The technique of x-ray fluorescence analysis has been used to make relative and absolute measurements of the amount of silver contained in x-ray films. These measurements are useful in evaluation of imaging properties such as speed and contrast and also in financial considerations associated with film costs and silver reclamation. The use of this technique will be described here and results will be presented from measurements on representative film samples. An attempt to apply the same method to analysis of the efficiency of spent fixer recovery systems is underway.

Several technical characteristics of digital fluorography are discussed. A derivation is made for the noise levels in the acquisition of digital fluorographic video images. The importance of high video signal-to-noise ratio is demonstrated and discussed. A noise analysis is made for reprocessing amplified difference images from analog video disk. It is shown that under the most stringent circumstances a disk with a signal-to-noise ratio of 200:1 increases the rms noise by at most 7%. Potential artifacts associated with the measurement of iodine sensitivity for digital radiographic systems are discussed. A phantom is presented which eliminates such artifacts. Experimental digital fluorographic phantom images are shown.

A set of computer models has been developed to generate images simulating those produced by projection x-ray systems. This was done to aid in the understanding of required sub-system performance and to study the implications of various system designs. For this purpose the basic processes of x-ray generation, attenuation and absorption were modeled. Scatter-to-primary ratios were modeled using published experimental data. A random number generator is used to model quantum as well as other sources of noise. These computer models agree well with experimental measurements. These image simulation programs have been used to predict the performance of digital radiographic systems (both scanned fan-beam and fluorographic). Images of both temporal and dual energy subtraction will be presented. The effects of x-ray spectrum shaping, scatter-to-primary ratio, QDE, veiling glare and electronic noise will be shown. The detection of simulated vessels containing low amounts of iodine by the various system alternatives will be compared.

The addition of digital image collection and storage to standard and newly developed x-ray imaging techniques has allowed spectacular improvements in some diagnostic procedures. There is no reason to expect that the developments in this area are yet complete. But no matter what further developments occur in this field, all the techniques will share a common element, digital image storage and processing. This common element alone determines some of the important imaging characteristics. These will be discussed using one system, the Medical MICRODOSE System as an example.

We have designed and built an experimental linear xenon detector array to test its suitability as an x-ray detector for digital radiography. This new design consists of only two plates separated by a 0.5 mm gap filled with high pressure xenon gas. One of the plates is a high voltage electrode and the other is a circuit board etched to give an array of metal collector strips focussed on the x-ray source. Each stria or detector element is 9.5 mm wide and 6.3 cm long. This detector is simpler and more economical to build than the CT xenon x-ray detectors, in which detector elements are separated by metal septa occupying approximately 40% of the detector cross-sectional area. This new xenon detector design is described along with its calculated quantum efficiency, and its measured sensitivity, linearity, and cross-talk between detector elements. From these results it is evident that this detector is suitable for digital radiography as it is dose efficient, and has less than 10% cross-talk.

The Quantex DS-20 digital image processor has been evaluated as a low cost alternative for digital vascular radiography. The processor is used in conjunction with a conventional remote radiographic/fluoroscopic system with no generator/processor interface required. (Use of a video-tape system reduces timing problems with minimal image degradation). Phantom studies indicate vessels as small as 1.5 mm can be seen with concentrations of contrast as low as 10 mgΓ/ml. Animal studies show that subtraction images of diagnostic quality can be generated.

Digital subtraction fluoroscopy is used to evaluate renal function. The system used in this paper consists of an x-ray intensifier-TV chain and a digital video image processor. Experimental results using dogs are presented.

A real-time digital fluoroscopic system utilizing two video processing units (Quantex) in tandem to produce continuous subtraction images of peripheral and internal vessels following intravenous contrast media injection has been inves-tigated. The first processor subtracts a mask image consisting of an exponentially weighted moving average of N1 frames (N1 = 2^k where k = 0.7) from each incoming video frame, divides by N1, and outputs the resulting difference image to the second processor. The second unit continuously averages N2 incoming frames (N2 = 2^k) and outputs to a video monitor and analog disc recorder. The contrast of the subtracted images can be manipulated by changing gain or by a non-linear output transform. After initial equipment adjustments, a subtraction sequence can be produced without operator interaction with the processors. Alternatively, the operator can freeze the mask and/or the subtracted output image at any time during the sequence. Raw data is preserved on a wide band video tape recorder permitting retrospective viewing of an injection sequence with different processor settings. The advantage of the tandem arrangement is that it has great flexibility in varying the duration and the time of both the mask and injection images thereby minimizing problems of registration between them. In addition, image noise is reduced by compiling video frames rather than by using a large radiation dose for a single frame, which requires a wide dynamic range video camera riot commonly available in diagnostic x-ray equipment. High quality subtraction images of arteries have been obtained in 15 anesthetized dogs using relatively low exposure rates (10-12 μR/video frame) modest volumes of contrast medium (0.5-1 ml/kg), and low injection flow rates (6-10 ml/sec). The results/ achieved so far suggest that this system has direct clinical applications.

Many quality control procedures in diagnostic radiology require the accurate simulation of the x-ray spectrum transmitted by the patient. Measurements of the contrast produced by mammographic imaging systems and the adjustment of phototimers used with rare earth screens are examples of such procedures. An analytical method for designing two-component phantoms which accurately simulate not only the narrow-beam attenuation but also the scattering properties of specified thicknesses of given tissue compositions is presented. As an example, a mammography phantom composition which matches both the narrow-beam attenuation and the scattering of 50% adipose/50% glandular breast tissue to better than ±0.5% at x-ray energies from 10 to 75 keV is described. An extension of the method to the design of three-component materials with improved simulation properties (±0.1%) is also described. These materials are suitable for tissue simulation in computed tomography (CT) applications.

One of the objectives of quality control in diagnostic radiography is the ability to select the technique factors which will produce a desired optical density on the final radiograph. This density is a function of several variables. Among these are x-ray target material, target angle, kV and the voltage wave form, mAs, geometry, patient thickness, field size, properties of the scatter reduction devices and type of screen film systems and processing employed. These variables have been modelled into a computer program. The program calculates technique factors of kV and mAs to give a desired film density for a given set of the other parameters. Application of the results of the theoretical calculations to experimental studies utilizing phantoms resulted in an optical density within 9% of the expected value. Results of the theoretical and experi-mental studies have been presented. Possible application of computer control in diagnostic radiography will be discussed.

A method has been developed to noninvasively measure the instantaneous potential (kV) applied to an x-ray tube. The method uses differentially filtered x-ray detectors whose outputs during the exposure are converted to digital signals and stored in a memory array. The conversions are made every 125 microseconds. After the exposure, a microprocessor calculates the ratios of the detector outputs; computes the kV waveform from stored calibration data; and digitally displays the kV value. A resolution of ±0.5 kV at 110 kV has been achieved.

An evaluation of phantom materials is presented based on attenuation coefficients and measurements of phantom transmission using an image-intensifier system. Thicknesses of acrylic and copper that are equivalent to various anatomical regions are presented along with a copper phantom that simulates the attenuation of several anatomical regions at specific kVp's. Tests for system-level performance of image intensifiers that make use of this phantom are briefly described.

The advent of the 14" large format triple mode image intensifier and the thin window dual mode image intensifier have made a significant impact on the basic phyrosophy regarding the utilization of imaging devices for some radiological procedures. The physical characteristics of these image intensifiers were sufficiently advanced to induce a different approach in the mind of radiologists to carry out radiological procedures that deviate from the conventional and common practices. For example, the 14" large format image intensifier may eliminate the need for a full-size cassette spot filming device. Instead, recording of images is accomplished with the image intensifier/100 mm photospot camera chain. Although this may not be a new idea, since there are numerous fluoroscopic systems that are equipped with 100 mm/105 mm photospot cameras. It is a new approach in the sense that when the 14" image intensifier is employed for fluoroscopy a full-size cassette spot filming device is eliminated entirely and only the 100 mm photospot camera is available for recording of images.

An important parameter in an image intensifier-based imaging system is the contrast of the image intensifier tube itself. This paper presents a photographic method for the measurement of the large-scale contrast of an image intensifier tube at the system level which can be performed in the clinical setting with equipment normally found in a modern Radiology Department. A strip of Lead is positioned on-center at the bottom of the grid so that a line image of 100% contrast will be presented to the input phosphor of the image intensifier when the x-ray tube is energized at low kilovoltage. The output phosphor is photographed either with an existing fluorographic camera (photospot or cine) on the imaging tower, or with a 35-mm SLR camera loaded with orthochromatic cine film through the collimating lens of the system from the position normally occupied by the television camera, during fluoroscopy, if no other camera is present on the imaging tower. The resultant on-frame optical density is measured with a densitometer in the central part of the developed frame adjacent to and behind the image of the Lead strip. These optical density readings are converted into a ratio of light intensities from the corresponding regions on the output phosphor using the characteristic curve for the type of film employed, which is obtained by means of light sensitometry. The percent contrast is then calculated from (ratio of intensities - 1)/(ratio of intensities + 1)) X 100. Using data for a variety of CsI image intensifiers used for both gastrointestinal and vascular studies, the method is shown to give measured percent contrasts with a reproducibility of no worse than-2%, independent of type of camera used or type of sensitometer used. Standards of acceptable performance based on the author's experience with this technique over the past five years are presented for systems designed for Barium studies and for systems designed for Iodine studies. The relationship between the percent contrast as defined here, the veiling glare as defined by Siedband, and the contrast ratio as normally defined is discussed, as well as the relevance of large-scale contrast of an image intensifier tube.

The measured equivalent x-ray tube focal spot size can be evaluated by either pinhole camera images or star resolution test object measurements. By a rather simple calculation using the measured equivalent focal spot size and the focal spot-to-screen film and object-to-screen film distances, the geometric resolution limit can be determined. This resolution limit can be used to indicate the needed focal spot size and focal spot-to-object distance for each diagnostic procedure to utilize properly the resolution capability of the screen-film system. This calculation is also important when screen-film systems of different reso-lution are considered for a particular diagnostic procedure. Data are presented for upper gastrointestinal, chest, breast, abdominal and extremity radiography.

This paper describes a computer program that is used to analyze the field uniformity of gamma cameras. The daily floods collected on the gamma camera are digitized and analyzed. The program calculates a number of quantities that pertain to the flood image as a whole and also the counts and the standard deviation in small areas centered over each photomultiplier tube. These daily results are compared either to a standard reference image or to the results averaged over the previous five days. Changes in the region of each photo-multiplier tube are displayed as an artificially constructed image. Observers find it easier to detect changes in this artificially constructed image than in the original image.

Computer studies of the three-dimensional imaging characteristics of a new single gamma emission tomography system have been conducted. In this system collimation is performed electronically by recording a sequential interaction of the gamma radiation with two detectors. A considerable gain in sensitivity is expected over a conventionally collimated system. Radiation is scattered from the first detector onto the second detector, and events are recorded in coincidence. The activity is thereby localized on conical surfaces inside the object to be imaged. A novel two-step iterative algorithm nas been developed to reconstruct the three-dimensional emission images for this system. Results of computer simulation studies are presented.

A computer generated mathematical model of a thallium-201 myocardial image is described which is based on realistic geometric and physiological assumptions. The left ventricle is represented by an ellipsoid truncated by aortic and mitral valve planes. Initially, an image of a motionless left ventricle is calculated with the location, size, and relative activity of perfusion defects selected by the designer. The calculation includes corrections for photon attenuation by overlying structures and the relative distribution of activity within the tissues. Motion of the ventricular walls is simulated either by a weighted sum of images at different stages in the cardiac cycle or by a blurring function whose width varies with position. Camera and collimator blurring are estimated by the MTF of the system measured at a representative depth in a phantom. Statistical noise is added using a Poisson random number generator. The usefulness of this model is due to two factors: the a priori characterization of location and extent of perfusion defects and the strong visual similarity of the images to actual clinical studies. These properties should permit systematic evaluation of image processing algorithms using this model. The principles employed in developing this cardiac image model can readily be applied to the simulation of other nuclear medicine studies and to other medical imaging modalities including computed tomography, ultrasound, and digital radiography.

To test if changing magnetic fields and RF fields used in NMR imagers could induce electrical currents in surgical clips and prostheses capable of causing localized tissue heating, we exposed steel surgical clips, copper wire clips, and hip prostheses to fields greater than those used in our NMR imager. Results indicate that no significant heating should be expected to occur from implanted surgical clips during exposure to our NMR imager. The heating of larger metallic implants should be further investigated.

A new class of diagnostic imaging systems, loosely termed Digital Radiography, has emerged during the past year from the laboratory into the forefront of radiological practice. These systems have in common the acquisition of a two dimensional projection image in digital format but include a variety of detectors, techniques and applications. Digital radiography apparatus are conveniently divided into point scanned, line scanned and area systems. Each system is briefly described and its performance capabilities are reviewed. Digital video radiography is given added emphasis due to its rapid development for intravenous angiography. The key requirements and present limitations of major components of the video system are noted. Gross specifications of the next generation of computers for digital radiography are postulated.

In July 1973, we became actively involved in a research program to establish the validity of photoelectronic digital radiology. Since that time our goal has been to ascertain whether state-of-the-art x-ray screens, intensifiers, optics, video cameras, and digital components can be combined into systems that will satisfactorily replace film in diagnostic radiology.

Silver halide is a unique imaging receptor because the absorption of only a few light photons can produce up to 10⁹ silver atoms. Recently, the cost of silver has increased dramatically although there is no shortage of the metal. These price increases have focused attention on making more efficient use of this precious metal in silver halide films and on its recovery after use. Photographic manufacturers have developed technology necessary to make X-ray films using less silver without affecting photographic performance. Other non-silver imaging systems have been developed for use when high photographic speed is not required.

Silver halides are the image-forming element in x-ray films. Silver used in x-ray emulsion either ends up as the black area of the film or is dissolved in the fixer solution during processing. Film density and film silver coverage determine how much silver is to be found in fixing solution or left on processed film. The choice between the use of electrolytic or metallic exchange silver equipment is determined by the amount of silver to be collected. The proper arrangement of the above two methods of silver recovery can produce high silver recovery efficiencies. Recovering silver from processed x-ray film is best left to companies which specialize in providing that service.

The use of information contained in the transmitted x-ray spectrum provides the capability for selective removal of substances of a particular mean atomic number in projection radiographs. Using a prototype system for line-scanned digital radiography, x-ray images were produced with two different x-ray spectra by modulating the kVp applied to the x-ray source and filtering the x-ray beam. Using a previously described Compton/photoelectric decomposition algorithm, subtraction images are made with bone or soft-tissue (water) shadows removed. Applications to chest, abdominal and skeletal imaging in-vivo are demonstrated.

This paper describes a new scheme of generating a sharply focused x-ray beam for obtaining radiopacity information at discretely sampled points on an object, optionally combined with a multiple detector system for further improving the radiopacity contrast. Zoom capabilities are provided for clinical diagnosis and other purposes. It is an application and further specialization of the x-ray microbeam system that was invented by one of the authors and implemented at the University of Tokyo for the purpose of acquiring data on articulatory movements [Fujimura et al., 1973]. A generalization of the scheme into a fan beam (rather than a pencil-beam) system, by the use of a slit in place of the pinhole, in combination with arrays of detectors, is also described, as well as a multiple pinhole system using more than one pencil beam simultaneously. Specially designed x-ray generating targets are proposed for these systems.

The quantitative nature of Computerized Tomography (CT) has suggested the use of the numbers to characterize tissues being imaged almost from the beginning of clinical application. Early attempts, however, quickly dampened the enthusiasm of some proponents even to the extent of leading them to state categorically that such applications are useless, if not downright misleading. There are several major stumbling blocks to overcome: 1. Calibration of the unit must be carefully maintained, 2. Beam hardening artifacts must be carefully accounted for, 3. Normal biological variation must be established as well as a listing of independent parameters that cause variations, 4. Common abnormalities must be catalogued as well as their influence on CT number, 5. Useable criteria to establish or suggest a given diagnosis must be identified and tested.

In this paper we will give a short review of the fundamental and practicle aspects of the use of contrast materials in CT with aim of augmenting detection and identification of various lesions of the brain. We will outline briefly some of the fundamental physical, chemical and pharmacological characteristics of current water soluble contrast materials. We will review some of the research that has been directed towards classification of the brain tumors with conventional contrast agents by a relative enhancement, image textural changes and slow dynamic effects (5-60 minutes).

The time course of dynamic contrast enhancement of brain tissue depends on the intrin sic tissue characteristics of blood volume, blood flow and capillary permeability, as well as on the nature of the contrast medium and the time course of the contrast input. In the normal brain, conventional iodinated contrast media are confined solely to the vascular space of the tissue, i.e., they are non-diffusible indicators. In abnormal brain tissue, iodinated contrast will diffuse into the extravascular space as well, which will affect the time course of the enhancement. Stable xenon may also be used as a contrast medium; it approximates a freely diffusible contrast material. Analysis of dynamic iodinated contrast enhancement in rapid sequence computed tomography images of arterial vessels and corresponding parenchymal areas permits approximate correction for the input function, and thus allows an approximate calculation of blood volume and blood flow in normal and ischemic brain. Simple compartment analysis may also permit some characterization of the extravascular exchange of iodinated contrast in more abnormal brain tissue. Measurement of the time course of exhaled xenon concentrations can permit similar calculations for dynamic xenon enhancement.

Several techniques for heavy-ion computerized tomography are being investigated at Lawrence Berkeley Laboratory. Using beams of carbon and neon from the Bevalac, we have demonstrated that these methods are feasible and capable of high resolution. We describe in some detail the method of heavy-ion CT imaging using nuclear track detectors, including a discussion of procedures for optical scanning and digitization of data and computerized distortion corrections. Comparisons between a heavy-ion CT image and X-ray CT image of a simple phantom are discussed. Preliminary results from two techniques using active, online detector systems for performing heavy-ion computerized tomography are presented. One method uses a multiplane, multiwire ionization chamber for detecting the heavy ions in a mode allowing true three-dimensional reconstructions. The other technique uses a system of position-sensitive silicon solid-state detectors for spacial information and high-purity germanium detectors to measure accurately the residual energy of the ions.

Split-filter CT is a simple technique for obtaining dual energy CT images after a single 360° x-ray source rotation. The method consists of creating low and high energy spectra from a single source fan by filtering the two halves of the fan differently. From a practical standpoint, the acquisition of high quality split-filter images from a commercial scanner is dependent upon proper source, detector and split-filter alignment as well as on proper spectral calibration. These problems are discussed and split-filter images corrected for alignment and calibration artefacts are shown. The noise performance of any dual energy scheme is related to the quality of spectral separation between low and high energy beams and to the efficiency of dose utilization. The theoretical noise performance of the split-filter method is compared to other dual energy techniques, namely the dual-kVp method, using a generalized dual energy noise analysis. Results of split-filter scans from a clinical scanner are presented to demonstrate the ease of obtaining Compton and photoelectric images and exact spectral artefact correction at arbitrary monoenergetic energies on a routine basis. Split-filter images resulting from subtraction of low and high energy measurements are also shown to illustrate enhanced visualization of iodine contrast.

Energy-selective computed tomography has several important properties useful for in-vivo tissue characterization. Most importantly, it produces more information than conventional computed tomography. This information can be considered to be an added dimension which can be used to eliminate the ambiguities in conventional CT data. The noise in energy-selective computed tomography is also two dimensional and an un-correlated coordinate system can be defined which is needed for studying the capabilities of the technique for characterizing tissues. By using the calibration material basis set, the information from energy-selective CT can be extracted with extreme accuracy. Our preliminary experiments indicate that the technique is accurate enough to characterize the difference between gray and white matter. Most conventional systems have difficulty in distinguishing these materials, much less characterizing the reason for their differing attenuation. Thus energy-selective CT has the promise of providing extremely accurate tissue characterization based on its physical properties.

The applicability of quantitative information contained in CT scans to diagnostic radiology and to radiation therapy treatment planning and the heterogeneity problem has been recognized by members of the radiological community and by manufacturers. Determination of the relationship between electron density and CT number is important for these applications. As CT technology has evolved, CT number generation has changed. CT number variation was limited in the early water bag systems. However, later generation "air" scanners may exhibit variation in CT numbers across a reconstructed image which are re-lated to positioning within the scan circle and scan field size. Results of experimental investigations using tissue-equivalent phantoms of different cross-sectional shapes and areas on the Technicare Delta 2020 are presented. Investigations also cover the effect of "shaped" and "flat" x-ray beam filters. A variation in CT number is demonstrated on this fan beam geometry scanner for phantoms of different sizes and for different scan circle diameters. An explanation of these effects is given. Differences of as much as 20% in determination of tissue electron density relative to water under different experimental conditions are obtained and reported. A family of curves (electron density vs. CT number) is presented for different patient cross-sectional areas and different scanner settings.

Four state of the art CT scanners were evaluated with respect to optimal techniques for multiplanar imaging. The four scanners were a G. E. 8800, Pfizer 0450, Picker Synerview 600 and Siemens Somatom 2. Patient movement artefacts can be minimized by choice of techniques that provide rapid data acquisition. By deferring reconstruction, using batch mode acquisition, suppressing screen display and operator interaction and minimizing tube loading as many as 33 thin slices can be acquired in < 8 minutes. This rapid scan technique makes use of the narrow collimation over a large (5.0 cm) distance quite reasonable. A high contrast star resolution phantom was scanned using these rapid scan techniques. The multiplanar images produced from narrow slices are much higher resolution than those created with more widely collimated slices. A low contrast (2.5%) resolution object scanned with the same rapid acquisition method shows improved resolution for the narrow collimation even in the presence of increased noise accompanying the narrow collimation.

The sampling geometry for conventional, multidetector, computed tomography is illustrated in terms of the Radon transformation for both rotate-rotate (3rd generation fan beam) and rotate-stationary (4th generation fan beam) scanners. By deriving an expression for the outline of the sampling region in the Radon transformation for each detector measurement it is demonstrated that the entire Radon transformation can be covered by non-overlapping sampling regions with the assumption of negligible detector dead space. An expression for angular aliasing is derived which demonstrates that object dependent artifacts can occur if the angular width of the sampling regions in conjunction with the angular sampling increment does not provide sufficient suppression of the high order angular harmonics in the representation of the scanned object. The number of views necessary to suppress angular aliasing, as well as the potential spatial resolution and general image quality are shown to be fundamentally related to the size, shape, and relative orientation of the Radon transformation sampling regions.

For a number of years multi-format or multi-image cameras have been used in radiology departments to record images for ultrasound, nuclear medicine and computerized tomography. The authors have described in previous papers the development of a total low dose fluoroscopy system, using a Siemens Videomed H 1023 line 25 MHz television system, a V.A.S. video disc recorder for pulsed fluoroscopy, and a modified Matrix Videoimager to record the spot film images. This approach has provided images of high quality with dose and cost reductions of the order of 90% for the total examination. The particular problems involved with the modification of a multi-image camera for fluoroscopic or radiographic procedures, can be minimised by appropriate choice of monitor phosphor and correct control of the exposure sequence.

Much of the information in real-time ultrasonic imaging examination is lost when only a few of the scan planes are recorded. Recording of all scan planes is particularly desirable in the examination of the breast, where there are few anatomical landmarks and discovery of small lesions leads to the best prognosis. We have developed a system that allows such recording and provides for playback in a very flexible manner: the images can be reviewed at a rate from zero (static) to 30 frames per second, forward or backward, with "instantaneous" reversal or stopping in the progression of images under review. The images are stored as two rings of 120 images each, one ring per breast, on a disc of photographic material. The developed disc is placed in a reader which spins it at a constant 60 rev/sec. By counting the timing marks recorded along with the images (obviating tight mechanical tolerances) any selected image can be flashed onto a TV camera by a strobe unit controlled by the timing-mark count. As there is no mechanical inertia, the rate of change in the selected image can proceed at any rate up to the monitor and is applicable for use with formatters, VTR, etc., for further image recording. The recording disc is small (8" dia-meter) and inexpensive (<$5) and provides an archival unit record for inclusion in patient files.

Patient x-ray exposure and operating room procedure time have been significantly reduced by replacing conventional film screen and fluoroscopic imaging techniques with an electronic imaging system. The electronic imaging system requires two mAs at 90 kvp to record an image as compared to 100 mAs at 90 kvp required by a rare earth film screen combination. An average of three minutes is required to expose and develop x-ray film while only 1/10 of a second is necessary to obtain an electronic image. Fluoroscopy is restricted in the O.R. because of high patient and personnel exposure.

Conventional x-ray tomograms are limited by the presence of defocussed contributions from over and underlying structures in the final image. Edges are relatively indistinct and image contrast low when compared to conventional radiographs of the same object. Assuming that the x-ray tomograph is linear and spatially invariant with a circularly symmetric blurring function, it is possible to compensate for the presence of defocus blur to a large extent. Conventional x-ray tomograms of a composite phantom (Littleton) were used to determine the system characteristics. Circularly and quadrilaterally symmetric blurring functions computed from the phantom images were used to calculate the blurred contribution of over and underlying structures. Processed images were obtained by subtracting the defocussed components of adjacent tomoplanes from the tomographic image obtained at the level of interest. This processing was performed using a PDP-11 minicomputer by discrete fixed point convolution operations with up to 19 by 19 operators applied to a 512 x 512 image matrix. The results demonstrate a substantial improvement in subjective image quality.

The compression of the digital information in a diagnostic image allows data to be stored in a smaller memory. This directly effects the cost and performance of computerized diagnostic imaging systems. Predictive compression (PC) combined with variable length coding (VLC) can achieve image compression with no alteration of the image element values. PC uses the values of neighboring, or previously transmitted, picture elements to predict a value for the next picture element. The difference between the prediction and the actual pixel value is then stored, in lieu of the actual pixel value. VLC techniques allow the differential signal to be encoded so that values which occur most frequently are stored with as little as one binary bit. The use of PC and VLC reduces the number of bits required to store an image. We have determined the compression factor (original/compressed bits per image) associated with five linear PC algorithms encoded with six VLC techniques for Ultrasound, Nuclear Medicine, Computed Tomography and Digital Roentgenographic images. The results allow us to identify PC and VLC algorithms most suitable for diagnostic imaging. Compression factors between 1.5 and 4 are achievable with simple algorithms.