Robert J. Jennings

A method for comparing beam-hardening filter materials for diagnostic radiology.

The necessity for using adequate beam filtration in diagnostic radiology is well known. Although aluminum is the most widely used filter material for diagnostic x-ray applications, the possibility that other materials might have superior properties has prompted a number of studies that have attempted to determine both the type and the amount of filtration most appropriate for a given situation. This paper describes a method based on precise matching of spectral shape that permits the absolute ranking of beam-hardening materials. Matching of spectral shape ensures equality of such parameters as image contrast and patient dose. Spectrally equivalent filters can then be ranked on the basis of the transmission of one relative to another. Following the development of the theory behind the method and an algorithm for implementing it, the method is applied to the evaluation of a variety of materials for use as filters in diagnostic radiology. Experimental verification of a few of the calculated results is also described. Both calculated and experimental results show that normal aluminum filters are about 10% less efficient than filters of materials such as copper, brass, or iron. Since the approach followed here was the basis for several early investigations of filtration for orthovoltage therapy, a brief comparison of results from these early reports with results calculated using the method developed here is also presented.

Beam quality independent attenuation phantom for estimating patient exposure from x-ray automatic exposure controlled chest examinations.

The periodic assessment of exposures in diagnostic radiology is an important part of a comprehensive quality assurance program. The most frequent radiologic examination conducted in the United States is chest radiography. Automatic exposure controlled (AEC) techniques are often used for this exam, and a standard patient-equivalent chest phantom is useful when estimating patient exposures on such systems. This is of particular importance if exposures are to be compared among AEC systems with different entrance x-ray spectra. Such a phantom has been developed to facilitate surveys of the average patient exposure from AEC posteroanterior chest radiography. The phantom is relatively lightweight and easily transportable, sturdy and made of readily available and relatively inexpensive materials (Lucite and aluminum). It accurately simulates the primary and scatter transmission through the lung-field regions of a patient-equivalent anthropomorphic phantom for x-ray spectra typically used in chest radiography. A clinical evaluation has been conducted to verify the patient equivalence of the phantom. Measurements of patient entrance skin exposure were obtained for a large number of patients on a variety of x-ray systems operated in the AEC mode using one or both lung-field detectors. Comparison of these data with exposure estimates derived from the phantom indicate that the phantom attenuates the x-ray beam in such a way that it can be employed to accurately and consistently estimate the mean exposure of the average patient under a variety of radiographic conditions. The design, development, and evaluation of the patient-equivalent attenuation phantom is described.

On the significance of very small angle scattered radiation to radiographic imaging at low energies

We have studied the angular distribution of scattered radiation at low energies for angles between 2 degrees and 15 degrees from the outward normal to the exit surface of several phantoms, with 1 degree resolution. A cryogenically cooled germanium detector was used to measure the spectra of the scattered radiation. The differential scattering fluences, or numbers of photons per unit solid angle per unit surface area, exhibit distinct peaks at angles in the vicinity of 5 degrees, with the angular position being only slightly energy dependent but very material dependent. The scattered spectra show large changes as a function of angle, in some cases actually becoming harder than the exit unscattered beam. The significance of this behavior relative to the imaging of targets in mammography examinations is discussed.

Anisotropic imaging performance in breast tomosynthesis.

We describe the anisotropy in imaging performance caused by oblique x-ray incidence in indirect detectors for breast tomosynthesis based on columnar scintillator screens. We use MANTIS, a freely available combined x-ray, electron, and optical Monte Carlo transport package which models the indirect detection processes in columnar screens, interaction by interaction. The code has been previously validated against published optical distributions. In this article, initial validation results are provided concerning the blur for particular designs of phosphor screens for which some details with respect to the columnar geometry are available from scanning electron microscopy. The polyenergetic x-ray spectrum utilized comes from a database of experimental data for three different anode/filter/kVp combinations: Mo/Mo at 28 kVp, Rh/Rh at 28 kVp, and W/Al at 42 kVp. The x-ray spectra were then filtered with breast tissue (3, 4, and 6 cm thickness), compression paddle, and support base, according to the oblique paths determined by the incidence angle. The composition of the breast tissue was 50%/50% adipose/glandular tissue mass ratio. Results are reported on the pulse-height statistics of the light output and on spatial blur, expressed as the response of the detector to a pencil beam with a certain incidence angle. Results suggest that the response is nonsymmetrical and that the resolution properties of a tomosynthesis system vary significantly with the angle of x-ray incidence. In contrast, it is found that the noise due to the variability in the number of light photons detected per primary x-ray interaction changes only a few percent. The anisotropy in the response is not less in screens with absorptive backings while the noise introduced by variations in the depth-dependent light output and optical transport is larger. The results suggest that anisotropic imaging performance across the detector area can be incorporated into reconstruction algorithms for improving the image quality of breast tomosynthesis. This study also demonstrates that the assessment of image quality of breast tomosynthesis systems requires a more complete description of the detector response beyond local, center measurements of resolution and noise that assume some degree of symmetry in the detector performance.

Noise in flat-panel displays with subpixel structure.

Subpixel structures found in medical monochrome active-matrix liquid crystal displays (AMLCDs) affect noise estimates measured with conventional methods. In this work, we discuss methods that identify sources of noise and permit the comparison of luminance noise estimates across technologies independent of pixel design and device technology. We used a three-million pixel AMLCD with a pixel structure consisting of three color stripes, each in a two-domain, in-plane switching mode. Images of uniform fields displayed on the AMLCD were acquired using a low-noise, high-resolution CCD camera. The camera noise and flat-field response were characterized using a uniform light source constructed for this purpose. We show results in terms of spatial luminance noise and noise power spectrum for high-resolution images and for the same images processed with a pixel-aligned aperture. We find that the pixel-aligned aperture eliminates almost all the noise found in the high-resolution images, suggesting that most of the luminance noise in AMLCDs comes from the subpixel structure and less-than-100% aperture ratio, rather than from interpixel variations.

A patient‐equivalent attenuation phantom for estimating patient exposures from automatic exposure controlled x‐ray examinations of the abdomen and lumbo–sacral spine

The Joint Commission on Accreditation of Healthcare Organizations requires diagnostic radiology facilities to known the approximate amount of radiation received by an average patient during radiographic examinations at the facility. Automatic exposure controlled (AEC) techniques are used for many of these exams, and a standard patient-equivalent phantom is necessary when estimating patient exposure on such systems. This is of particular importance if exposures are to be compared among AEC systems with different entrance x-ray spectra. We have developed a phantom, LucA1 Abdomen, to facilitate determining the average patient exposure from AEC anteroposterior (AP) abdomen and lumbo-sacral (LS) spine radiography. The phantom is relatively lightweight, transportable, sturdy, and made of readily available inexpensive materials (Lucite and aluminum). It accurately simulates the primary and scatter transmission through the soft tissue and L-4 spinal regions of a patient-equivalent anthropomorphic phantom for x-ray spectra typically used in abdomen/LS spine radiography. A clinical evaluation to verify the patient-equivalence of three commercial anthropomorphic phantoms (Humanoid, Rando, 3-M) and two acrylic/aluminum phantoms (ANSI and LucA1 Abdomen) has been conducted. The design and development of the LucA1 Abdomen phantom and the evaluation of all phantoms is described.

Evaluation of x-ray sources for mammography

A computational approach is being developed for the evaluation of mammographic imaging system performance. This approach takes into account both the spatial frequency properties and the x-ray spectral characteristics of the system being evaluated. The initial version of the program that implements the approach has been used to evaluate a conventional mammography source assembly for several breast thicknesses, and to compare the conventional tube and filter combination to alternatives that have been suggested for the imaging of breasts that are thicker or more dense than average. It has also been used to study the effect of varying the thickness of the molybdenum filter in the conventional system. The parameters calculated include contrast, average glandular dose, tube load, and a figure of merit, SNR2/Dose. The calculations confirm the strong dependence of system performance on both tube potential and breast thickness for the standard system, and indicate that alternative designs can improve performance in the imaging of thicker or more dense breasts. The study of filter thickness shows that, of the four parameters calculated, only tube load is strongly affected by filter thickness.

An approach to specifying a minimum dose system for mammography using multiparameter optimization techniques

Analytical expressions have been written for image quality in mammography. Multiparameter optimizations have been conducted to find mammography systems providing the lowest patient dose for a given image quality. The optimizations are subject to constraints imposed by technology, such as power limits on the tube focal spot, absorption efficiency related to detector resolution, and others. The optimizations permit system geometry, kVp, filtration, detector resolution, focal spot size, and grid characteristics to vary simultaneously and self-consistently subject to the constraints. A system configuration approaching a factor of 3 dose reduction has been found without assuming radical technological advances. The system satisfies image quality constraints for both large and small targets and would be possible to implement clinically. The sensitivity of the results to the assumptions made in the modeling has been investigated.

Comparison of beam‐hardening and K‐edge filters for imaging barium and iodine during fluoroscopy

This study investigated the dose reduction performance of several beam-hardening and K-edge filter materials for the imaging of barium or iodine during fluoroscopy. A computer model was developed to simulate the effect of added filtration on entrance exposure rate (Xp), integral dose rate (Di), contrast (C), signal to noise ratio (SNR), imaging performance per dose (SNR2/Di), and tube load. The model incorporated the response characteristics, in both manual and automatic control modes of operation, of fluoroscopic systems to increasing or decreasing x-ray intensity at the input of the image intensifier. Input parameters to the computer model included choice of filter material and thickness, a barium or iodine test object, tube potential, phantom thickness, a CsI input phosphor, and a set of algorithms for controlling the fluoroscopic system. In all cases, the performance of systems with added filtration was judged with respect to a reference system operating under comparable conditions. In general, either beam-hardening or K-edge filters provided a significant reduction in entrance exposure and integral dose rates, but with an attendant increase in tube load. For a fluoroscopic system constrained to follow a representative automatic brightness control algorithm, added filtration provided a reduction in entrance exposure and integral dose rates for all phantom or uniformly distributed barium thickness. However, the imaging performance per dose, in some cases, decreased rapidly and was less than that of the reference system at large thicknesses. Only as change in the algorithm controlling the kVcp and mA operating points on the fluoroscopic system provided an imaging performance per dose greater than the reference systems at large thicknesses. The practical implementation of adding filtration to fluoroscopic systems is most simply accomplished with beam-hardening filters rather than K-edge filters. However, the systems with K-edge added filtration can provide slightly better performance when used over a limited range of phantom thicknesses such as the range normally associated with pediatric patients.

Noise analysis of scintillation camera images: stochastic and non-stochastic effects

An analysis of noise was performed on scintillation camera images. The analysis was performed on sets of field floods, uniform exposures over the field of view, in such a way that stochastic and non-stochastic effects were separable. Stochastic noise for the digitally acquired images was found to approach the limit of Poisson count statistics. The non-stochastic component of the calculations dominated the results at the lower spatial frequencies. These results hold true even after standard methods of uniformity correction are applied, primarily due to the effects of Compton scatter events. The noise power spectrum calculations carry information about the patterns of the noise in flood images. Examples of digitisation, photomultiplier tube mottle and edge-packing artefacts are presented.