Gary T. Barnes

Semiempirical model for generating tungsten target x-ray spectra

A semiempirical model for generating tungsten target x-ray spectra is presented. This model extends earlier work in two significant areas. First, both bremsstrahlung and characteristic x-ray production are assumed to occur at varying depths within the target. Second, optimal parameters for the model were determined from experimental spectra utilizing nonlinear least-squares techniques. As a result, good agreement is obtained between calculated and measured x-ray tube spectra and output for different target angles and a wide range of x-ray tube potentials. Such is not the case with previously published models.

Molybdenum target x-ray spectra: a semiempirical model.

A semiempirical model for generating molybdenum target x-ray spectra is presented. The model is an extension of a previous model developed by the authors for tungsten and takes into account the depth of production for both bremsstrahlung and characteristic x-ray photons. As in the previous work, the optimal model parameters were determined using nonlinear least-squares fits to experimental data. Good agreement between the two was obtained. By varying target angle, off-axis angle, and filtration in the model in accordance with the x-ray tube and geometry of interest, results consistent with tabulated spectra for different conditions have been obtained.

An energy sensitive cassette for dual-energy mammography

A cassette for simultaneously acquiring dual-energy mammographic images is proposed and studied utilizing a theoretical noise analysis model. The cassette consists of a sandwich of two storage phosphor plates separated by a copper filter. The front, low atomic number, plate was assumed to be comprised of SrFBr and the rear, high atomic number, plate of commercially available BaFBr (98 mg/cm2 coating weight). Assuming a constant x-ray tube voltage of 50 kVp and a typical breast thickness, the theoretical model yielded a front SrFBr phosphor coating weight of approximately 21 mg/cm2. The study indicates that a relatively large separation in the average of x-ray photon energies absorbed in the two plates can be obtained. It also indicates that both a high-quality conventional (single energy) digital image and a tissue canceled digital image (i.e., a calcium image) can be obtained at dose levels comparable to those currently employed. The latter image could potentially improve the early detection of cancerous microcalcifications and also lends itself to computer aided diagnosis.

X-RAY TUBE POTENTIAL, FILTRATION, AND DETECTOR CONSIDERATIONS IN DUAL-ENERGY CHEST RADIOGRAPHY

The effect of x-ray tube potential and prepatient and interdetector filtration in single exposure dual energy chest imaging has been studied employing a carefully benchmarked model. The analysis utilized published methodology. Noise in simulated lung and mediastinum fields of the aluminum (bone) and Lucite (soft tissue) images were studied at fixed entrance skin exposure (ESE) for commonly employed sandwich detector and sandwich imaging plate configurations. Our results indicate noise in the lung increases slowly with tube potential above 120 kVp, while noise in the mediastinum decreases rapidly. Also, at high tube potential (> or = 120 kVp) adding moderate amounts of prepatient K-edge filtration (approximately equal to 100 mg/cm2) while optimizing imaging conditions for the lung tends to decrease noise in the lungs by approximately equal to 30% while increasing noise in the mediastinum by a similar amount. Without K-edge prepatient filtration, noise in the lung is minimized with Cu interdetector filter weights near 400 mg/cm2. In the mediastinum noise is minimized with heavier interdetector filter and prepatient K-edge filter weights. Prepatient K-edge filter weights that minimize image noise in either field can increase the tube loading by factors ranging from 10 to 10(10). Systems designed with sandwich detectors using commercially available phosphors and coating weights can produce contrast-to-noise ratios (CNRs) as high as 50% of the theoretical limit (defined as an optimized system with a totally absorbing rear detector).

A method to measure the MTF of digital x‐ray systems

A method has been devised to accurately measure the modulation transfer function (MTF) of digital x-ray systems up to and, for undersampled systems, beyond the pixel Nyquist frequency (fN). A phantom consisting of an array of parallel tungsten or similar wires is imaged, and discrete Fourier transforms of rows of pixel values are computed. Under suitable conditions of phantom orientation, wire diameter, wire spacing, and image magnification, the envelope of the modulus of the mean Fourier transform represents the system MTF. Experimental results extending beyond fN are presented for an undersampled prototype digital chest x-ray system and shown to be in reasonable agreement with predicted values. Employment of the method with other digital imaging modalities [i.e., computerized tomography (CT) scanners and nuclear magnetic resonance (NMR) units] is also discussed as well as error considerations and practical problems in implementing the method.

Bone mineral densitometry with x-ray and radionuclide sources : a theoretical comparison

Two methods of dual-photon absorptiometry (DPA) utilizing an x-ray tube instead of a radionuclide source have recently been introduced. In one method kVp switching is employed and two transmitted intensities at each pixel are determined. In the other method, K-edge filtration combined with a single kVp spectrum is used, but photons in two energy windows are counted. We present a theoretical analysis of the two methods, focusing on a figure of merit which is essentially the exposure efficiency (the precision for a given entrance exposure) and tube loading. We also compare their exposure efficiencies to theoretical limits that no DPA system can exceed. Our study indicates that the K-edge-filtered method is more exposure efficient by about a factor of 2. The switched-kVp method requires less heat units per scan by about a factor of 3. A hybrid K-edge switched-kVp method is suggested which achieves the same exposure efficiency as the K-edge-filtered method at lesser tube loading. Our theoretical model is based on published x-ray spectra and attenuation coefficients and is in good agreement with other simulation work. It is of interest that a point source of Gd-153 would be even more exposure efficient, achieving about 90% of the theoretical limit. However, in practice, the Gd source is of finite size and limited strength, and consequently the radionuclide method cannot achieve as good a precision as either x-ray method in similar scan times.

Image processing in digital radiography: Basic concepts and applications

Digital x-ray images are routinely processed to enhance diagnostic information and to suppress irrelevant detail, and also to extract quantitative information. The basic concepts and terminology of image processing as it applies to x-ray projection radiography are discussed and defined. In general, the processing of an image involves one or more point, local, and/or global operations. Clinical examples of linear and nonlinear gray-scale and algebraic point operations are presented. Examples are also given of local operations. Included in the latter group are distortion corrections, misregistration corrections, linear filtering, and nonlinear filtering.

Characterization of the reciprocity law failure in three mammography screen–film systems

The purpose of this project was to quantify reciprocity law failure (RLF) for mammography screen-film systems. Three widely used screen-film systems were evaluated: the Kodak MinR 2000 system. Fuji UM Mammo Fine screen and Fuji UM MA HC film, and Agfa MR Detail screen and Agfa Mammoray MR5 film. The logit algorithm that linearizes logistic curve shapes was utilized to characterize film sensitometric response. Different values of mammographic phantom thickness, tube current, and kVp were used to vary screen-film exposure rates. RLF was quantified by examining the dependence of logit parameters (maximum and minimum film density, curve shift, and slope) on exposure rate. The shift of the logit curve was found to be a good indicator of the screen-film system speed, while the slope of the logit curve is affected by the RLF. RLF leads to changes in film contrast as well as speed. For the range of exposure rates measured (50-fold), screen-film contrast and speed varied by factors of 2 and 3.5, respectively. Film contrast decreased as exposure rate increased. The greatest changes were observed with the Kodak MinR 2000 screen-film system.

Slit camera focal spot measurement errors in mammography.

Mammography x-ray tube focal spot sizes are routinely measured during acceptance testing and annual performance audits. The National Electrical Manufactures Association (NEMA) recommends the slit camera for this purpose. Investigated were the effect of slit rotational misalignment, tilt misalignment, image film density, film and screen-film image receptors, microscope magnification and reticule accuracy, and observer variation on slit camera focal spot measurements. Our results indicate that small rotational misalignment (< 5 degrees) and tilt misalignment (< 3 degrees) introduce insignificant error. Measured focal spot size increased slightly with image optical density, indicating that for consistent results the image optical density variations should be minimized. Also desirable for accurate field measurements is a high power microscope (25-50x) and a reticule with divisions of < or = 0.02 mm. Screen-film imaging consistently resulted in a slightly smaller measured focal spot size than direct film. The greatest source of error was due to observer variation. Of interest is that reader variability showed a consistent pattern and variation between two measurements by the same observer was much smaller than between observer variation, suggesting that standardized criteria should be established and a method of reader training developed. The length of the focal spot is defined at a reference axis angle specified by the mammography unit manufacturer. Presented is a tabulation of the focal spot geometry and reference axis angles for the majority of mammography units currently and recently marketed in North America.

An examination of errors in characteristic curve measurements of radiographic screen/film systems.

The precision and accuracy achieved in the measurement of characteristic curves for radiographic screen/film systems is quantitatively investigated for three techniques: inverse square, kVp bootstrap, and step-wedge bootstrap. Precision of all techniques is generally better than +/- 1.5% while the agreement among all intensity-scale techniques is better than 2% over the useful exposure latitude. However, the accuracy of the sensitometry will depend on several factors, including linearity and energy dependence of the calibration instrument, that may introduce larger errors. Comparisons of time-scale and intensity-scale methods are made and a means of measuring reciprocity law failure is demonstrated.