Ian A. Cunningham

Empirical and theoretical investigation of the noise performance of indirect detection, active matrix flat‐panel imagers (AMFPIs) for diagnostic radiology

Noise properties of active matrix, flat-panel imagers under conditions relevant to diagnostic radiology are investigated. These studies focus on imagers based upon arrays with pixels incorporating a discrete photodiode coupled to a thin-film transistor, both fabricated from hydrogenated amorphous silicon. These optically sensitive arrays are operated with an overlying x-ray converter to allow indirect detection of incident x rays. External electronics, including gate driver circuits and preamplification circuits, are also required to operate the arrays. A theoretical model describing the signal and noise transfer properties of the imagers under conditions relevant to diagnostic radiography, fluoroscopy, and mammography is developed. This frequency-dependent model is based upon a cascaded systems analysis wherein the imager is conceptually divided into a series of stages having intrinsic gain and spreading properties. Predictions from the model are compared with x-ray sensitivity and noise measurements obtained from individual pixels from an imager with a pixel format of 1536 x 1920 pixels at a pixel pitch of 127 microns. The model is shown to be in excellent agreement with measurements obtained with diagnostic x rays using various phosphor screens. The model is used to explore the potential performance of existing and hypothetical imagers for application in radiography, fluoroscopy, and mammography as a function of exposure, additive noise, and fill factor. These theoretical predictions suggest that imagers of this general design incorporating a CsI: Tl intensifying screen can be optimized to provide detective quantum efficiency (DQE) superior to existing screen-film and storage phosphor systems for general radiography and mammography. For fluoroscopy, the model predicts that with further optimization of a-Si:H imagers, DQE performance approaching that of the best x-ray image intensifier systems may be possible. The results of this analysis suggest strategies for future improvements of this imaging technology.

Signal, noise power spectrum, and detective quantum efficiency of indirect-detection flat-panel imagers for diagnostic radiology

The performance of an indirect-detection, active matrix flat-panel imager (FPI) at diagnostic energies is reported in terms of measured and theoretical signal size, noise power spectrum (NPS), and detective quantum efficiency (DQE). Based upon a 1536 x 1920 pixel, 127 microns pitch array of a-Si:H thin-film transistors and photodiodes, the FPI was developed as a prototype for examination of the potential of flat-panel technology in diagnostic x-ray imaging. The signal size per unit exposure (x-ray sensitivity) was measured for the FPI incorporating five commercially available Gd2O2S:Tb converting screens at energies 70-120 kVp. One-dimensional and two-dimensional NPS and DQE were measured for the FPI incorporating three such converters and as a function of the incident exposure. The measurements support the hypothesis that FPIs have significant potential for application in diagnostic radiology. A cascaded systems model that has shown good agreement with measured individual pixel signal and noise properties is employed to describe the performance of various FPI designs and configurations under a variety of diagnostic imaging conditions. Theoretical x-ray sensitivity, NPS, and DQE are compared to empirical results, and good agreement is observed in each case. The model is used to describe the potential performance of FPIs incorporating a recently developed, enhanced array that is commercially available and has been proposed for testing and application in diagnostic radiography and fluoroscopy. Under conditions corresponding to chest radiography, the analysis suggests that such systems can potentially meet or even exceed the DQE performance of existing technology, such as screen-film and storage phosphor systems; however, under conditions corresponding to general fluoroscopy, the typical exposure per frame is such that the DQE is limited by the total system gain and additive electronic noise. The cascaded systems analysis provides a valuable means of identifying the limiting stages of the imaging system, a tool for system optimization, and a guide for developing strategies of FPI design for various imaging applications.

Signal-to-noise optimization of medical imaging systems

Over recent decades a quiet revolution has taken place in the application of modern imaging theory to many fields of applied imaging. Nowhere has this movement been more dramatic than within the field of diagnostic medical x-ray imaging, to the extent that there is now a growing consensus around a universal imaging language for the description and comparison of the increasingly diverse range of technologies. This common language, which owes much to the basic quantum-limited approach pioneered by Rose and his contemporaries, embodies the fundamentally statistical nature of image signals and enables scientists and engineers to simultaneously develop new system designs optimized for the detection of small signals while constraining patient x-ray exposures to tolerable levels. We attempt to provide a summary of some of the more salient features of progress being made in the understanding of the signal-to-noise limitations of medical imaging systems and to place this progress within a historical context. Emphasis is placed on medical diagnostics based on x-ray imaging techniques.

A method for modulation transfer function determination from edge profiles with correction for finite-element differentiation.

In this paper we describe a technique for determining the modulation transfer function (MTF) of an imaging system from an experimentally obtained edge profile. The technique includes an exact correction for the frequency passband of the finite-element differentiation required to obtain the line spread function from the edge spread function. This correction has been ignored by investigators in the past and is required whenever finite-element differentiation is used rather than analytic differentiation of a model fitted to the edge response data. The magnitude of the MTF correction is approximately 11% at f = fc/2 and approximately 57% at f = fc, where fc = fs/2 is the maximum frequency reproducible without aliasing with a sampling rate of fs. The correction is performed in the spatial frequency domain by multiplying the uncorrected MTF by 1/sinc (pi f/2fc). A computer simulation is presented to demonstrate the effect and the correction procedure. An experimental MTF of an x-ray image intensifier system obtained using this technique is found to be consistent with an MTF obtained using a bar pattern test phantom.

Signal and noise in modulation transfer function determinations using the slit, wire, and edge techniques

The modulation transfer function (MTF) of an idealized imaging system can be determined from the Fourier transform of the systems line-spread function (LSF). Three techniques of experimentally determining the LSF require imaging either a slit, wire, or edge. In this paper, these three techniques are modeled theoretically to determine the noise in the calculated MTFs as a function of spatial frequency resulting from both quantum fluctuations and stochastic detector noise. The techniques are compared using the signal-to-noise ratio (SNR) in the MTF, defined as the ratio of the MTF value to the standard deviation in an ensemble of MTF determinations from independent measurements. It is shown that for a specified photon fluence, the edge method MTF has the highest SNR at low spatial frequencies, while that of the slit method is superior at high frequencies. The wire method SNR is always inferior to that of the slit technique. This suggests that the edge method is preferable for measuring parameters such as the low-frequency drop, and the slit method is preferable for determining high-frequency response. The cross-over frequency at which the slit and edge methods are equal (f(e)) for quantum-noise limited systems is a function of the slit width and the length over which the LSF is measured. For detector-noise limited systems, f(e) is dependent on the slit width only. The SNR in all but the quantum-noise limited slit method can therefore be increased by decreasing the length over which the LSF is measured, smoothing the tails of the LSF, or by fitting the tails to an analytic expression.

Parallel cascades: New ways to describe noise transfer in medical imaging systems

A generalized approach to describing transfer of the noise power spectrum through medical imaging systems has been developed over the past several years in which image-forming processes are represented in terms of a cascade of amplified point processes. Until recently, this approach has been restricted to serial cascades only. Here we develop a generalized expression for the cross covariance of amplified point processes and an expression for the cross spectral density for wide-sense stationary conditions. These results extend the generalized transfer-theory approach to include the description of more complex image-forming processes involving parallel cascades of quantum amplification processes. This parallel-cascade approach is used to develop a theoretical expression for noise-power transfer in a simple radiographic screen that includes the effect of characteristic x-ray reabsorption. The result confirms earlier work by Metz and Vyborny, who showed that reabsorption increases image noise and decreases the detective quantum efficiency at low spatial frequencies. Use of the transfer-theory approach facilitates a straightforward generalization to many new digital imaging systems including conventional angiographic and active-matrix flat-panel systems.

Segmented crystalline scintillators: An initial investigation of high quantum efficiency detectors for megavoltage x‐ray imaging

Electronic portal imaging devices (EPIDs) based on indirect detection, active matrix flat panel imagers (AMFPIs) have become the technology of choice for geometric verification of patient localization and dose delivery in external beam radiotherapy. However, current AMFPI EPIDs, which are based on powdered-phosphor screens, make use of only approximately 2% of the incident radiation, thus severely limiting their imaging performance as quantified by the detective quantum efficiency (DQE) (approximately 1%, compared to approximately 75% for kilovoltage AMFPIs). With the rapidly increasing adoption of image-guided techniques in virtually every aspect of radiotherapy, there exist strong incentives to develop high-DQE megavoltage x-ray imagers, capable of providing soft-tissue contrast at very low doses in megavoltage tomographic and, potentially, projection imaging. In this work we present a systematic theoretical and preliminary empirical evaluation of a promising, high-quantum-efficiency, megavoltage x-ray detector design based on a two-dimensional matrix of thick, optically isolated, crystalline scintillator elements. The detector is coupled with an indirect detection-based active matrix array, with the center-to-center spacing of the crystalline elements chosen to match the pitch of the underlying array pixels. Such a design enables the utilization of a significantly larger fraction of the incident radiation (up to 80% for a 6 MV beam), through increases in the thickness of the crystalline elements, without loss of spatial resolution due to the spread of optical photons. Radiation damage studies were performed on test samples of two candidate scintillator materials, CsI(Tl) and BGO, under conditions relevant to radiotherapy imaging. A detailed Monte Carlo-based study was performed in order to examine the signal, spatial spreading, and noise properties of the absorbed energy for several segmented detector configurations. Parameters studied included scintillator material, septal wall material, detector thickness, and the thickness of the septal walls. The results of the Monte Carlo simulations were used to estimate the upper limits of the modulation transfer function, noise power spectrum and the DQE for a select number of configurations. An exploratory, small-area prototype segmented detector was fabricated by infusing crystalline CsI(Tl) in a 2 mm thick tungsten matrix, and the signal response was measured under radiotherapy imaging conditions. Results from the radiation damage studies showed that both CsI(Tl) and BGO exhibited less than approximately 15% reduction in light output after 2500 cGy equivalent dose. The prototype CsI(Tl) segmented detector exhibited high uniformity, but a lower-than-expected magnitude of signal response. Finally, results from Monte Carlo studies strongly indicate that high scintillator-fill-factor configurations, incorporating high-density scintillator and septal wall materials, could achieve up to 50 times higher DQE compared to current AMFPI EPIDs.

Optimal phosphor thickness for portal imaging

A theoretical approach known as quantum accounting diagram (QAD) analysis has been used to calculate the spatial-frequency-dependent detective quantum efficiency (DQE) of two portal imaging systems: one based on a video camera and another based on an amorphous silicon array. The spatial frequency-dependent DQEs have then been used to determine indices of displayed and perceived image quality. These indices are figures of merit that can be used to optimize the design of linear imaging systems. We have used this approach to determine which of eight phosphor screen thicknesses (ranging between 67 and 947 mg/cm2) is optimal for the two designs of portal imaging systems. The physical characteristics (i.e., detection efficiencies, gains, and MTFs) of each of the eight x-ray detectors have been measured and combined with the physical characteristics of the remaining components to calculate the theoretical DQEs. In turn, the DQEs have been used to calculate theoretical indices of displayed and perceived image quality for two types of objects: a pelvis object and a pointlike object. The maximal indices of displayed and perceived image quality were obtained with screen thickness ranging between 358 and 947 mg/cm2, depending upon the imaging system design and the object being imaged. Importantly, the results showed that there is no single optimal screen thickness. The optimal thickness depended upon imaging task (e.g., detecting large, low-contrast structures, or detecting edges and small structures). Nevertheless, the results showed that there were only modest improvements in the indices of image quality for phosphor screens thicker than 350-400 mg/cm2.

The role of x-ray Swank factor in energy-resolving photon-counting imaging

PURPOSE Energy-resolved x-ray imaging has the potential to improve contrast-to-noise ratio by measuring the energy of each interacting photon and applying optimal weighting factors. The success of energy-resolving photon-counting (EPC) detectors relies on the ability of an x-ray detector to accurately measure the energy of each interacting photon. However, the escape of characteristic emissions and Compton scatter degrades spectral information. This article makes the theoretical connection between accuracy and imprecision in energy measurements with the x-ray Swank factor for a-Se, Si, CdZnTe, and HgI2-based detectors. METHODS For a detector that implements adaptive binning to sum all elements in which x-ray energy is deposited for a single interaction, energy imprecision is shown to depend on the Swank factor for a large element with x rays incident at the center. The response function for each converter material is determined using Monte Carlo methods and used to determine energy accuracy, Swank factor, and relative energy imprecision in photon-energy measurements. RESULTS For each material, at energies below the respective K edges, accuracy is close to unity and imprecision is only a few percent. Above the K-edge energies, characteristic emission results in a drop in accuracy and precision that depends on escape probability. In Si, and to some extent a-Se, Compton-scatter escape also degrades energy precision with increasing energy. The influence of converter thickness on energy accuracy and imprecision is modest for low-Z materials but becomes important when using high-Z materials at energies greater than the K-edge energies. CONCLUSIONS Accuracy and precision in energy measurements by EPC detectors are determined largely by the energy-dependent x-ray Swank factor. Modest decreases in the Swank factor (5%-15%) result in large increases in relative imprecision (30%-40%).

Tomographic imaging of the angular-dependent coherent-scatter cross section

A new special-purpose computed tomographic (CT) imaging system is described which produces images based on measurements of the low-angle (0-10 degrees) x-ray diffraction properties of an object. Low-angle scatter in the diagnostic x-ray energy range is dominated by coherent scatter, and the system uses first-generation CT geometry to acquire a diffraction pattern for each pencil beam. The patterns are used to reconstruct a series of images which represent the coherent-scatter intensity at a series of scatter angles. To demonstrate the potential of coherent-scatter CT (CSCT), the scanner has been built and used to image a phantom consisting of a water-filled Lucite cylinder containing rods of polyethylene, Lucite, polycarbonate, and nylon. In this paper, the system is described and a sequence of CSCT images of this phantom is shown. Coherent-scatter cross sections of these materials are generated for each pixel from this sequence of images and compared with cross sections measured separately. The resulting excellent agreement shows that the angular-dependent coherent-scatter cross section can be accurately imaged in a tomographic slice through an object. These cross sections give material-specific information about the object. The long-term goal of this research is to make measurements of bone-mineral content for every pixel in a tomographic slice.