IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 1819 Analytic Model of Energy-Absorption Response Functions in Compound X-ray Detector Materials Seungman Yun, Ho Kyung Kim*, Hanbean Youn, Jesse Tanguay, and Ian A. Cunningham Abstract—The absorbed energy distribution (AED) in X-ray imaging detectors is an important factor that affects both energy resolution and image quality through the Swank factor and detective quantum efficiency. In the diagnostic energy range (20–140 keV), escape of characteristic photons following photoelectric absorption and Compton scatter photons are primary sources of absorbed-energy dispersion in X-ray detectors. In this paper, we describe the development of an analytic model of the AED in compound X-ray detector materials, based on the cascaded-systems approach, that includes the effects of escape and reabsorption of characteristic and Compton-scatter photons. We derive analytic expressions for both semi-infinite slab and pixel geometries and validate our approach by Monte Carlo simulations. The analytic model provides the energy-dependent X-ray response function of arbitrary compound materials without time-consuming Monte Carlo simulations. We believe this model will be useful for correcting spectral distortion artifacts commonly observed in photon-counting applications and optimal design and development of novel X-ray detectors. or electron-hole pairs in a photoconductor, and liberation of these secondary quanta is determined by the random processes of X-ray energy deposition [1]. For example, at diagnostic energies (20–140 keV), variations in absorbed energy due to random escape of characteristic photons following a photoelectric interaction (PE) or scatter photons following a Compton (incoherent) interaction are primary causes of absorbed energy dispersion [2], [3] that increase variability in deposited energy in both energy-integrating [4] and photon-counting X-ray systems [3], [5]. Accurate and precise measurements of incident photon energy are particularly important for photon-counting techniques such as K-edge imaging [6]. Over the past several decades, many researchers [4], [7], [3], [8], [9] have characterized the effects of energy dispersion on X-ray image quality in terms of the absorbed energy distribution (AED) [keV ] given by Index Terms—Absorbed energy distribution (AED), compound semiconductor, Compton scattering, detector response function, digital radiography, fluorescence, photoelectric absorption, photon-counting imaging, X-ray convertor. (1) I. INTRODUCTION T HE SIGNAL generated by X-ray detectors is normally proportional to the charge collected by each element in an array of detector elements. This charge is proportional to the number of secondary image quanta liberated in the X-ray convertor, which may be optical photons in a phosphor Manuscript received April 04, 2013; revised April 30, 2013; accepted May 24, 2013. Date of publication June 03, 2013; date of current version October 02, 2013. This work was supported by the Basic Science Research Program through the National Research Foundation (NRF) of Korea funded by the Ministry of Education, Science and Technology (2011-0009769), the Canadian Institutes of Health Research (CIHR) operating grant and Natural Sciences and Engineering Research Council (NSERC) discovery grant (386330-2010). Asterisk indicates corresponding author. S. Yun is with the Mechanical Engineering Department, Pusan National University, Busan 609-735, Korea, and also with Imaging Research Laboratories, Robarts Research Institute, London, ON, N6A 5K8 Canada (e-mail: [email protected]). *H. K. Kim is with the Mechanical Engineering Department and the Center for Advanced Medical Engineering Research, Pusan National University, Busan 609-735, Korea (e-mail: [email protected]). H. Youn is with the Mechanical Engineering Department, Pusan National University, Busan 609-735, Korea (e-mail: [email protected]). J. Tanguay, and I. A. Cunningham are with Imaging Research Laboratories, Robarts Research Institute, and Department of Medical Biophysics, University of Western Ontario, London, ON, N6A 5K8 Canada (e-mail: jessetan@robarts. ca; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TMI.2013.2265806 where represents the incident X-ray photon energy, [keV ] describes the spectral distribution of incident photons, [keV ] represents the absorbed X-ray energy, and is the energy response function describing the average spectral given an incident photon with density of absorbed energy . energy Experimental determinations of the AED, such as the work of Blevis et al. [10], provide performance information of both conventional and photon-counting detectors. Alternatively, Monte Carlo (MC) methods can be used to accommodate complex geometries but generally require substantial computational effort. Analytic models are attractive as they can provide more insight into the performance limitations of these systems. For example, LeClair et al. [11] described the AED of a CdZnTe spectroscopy detector but did not consider the effects of X-ray scatter. Scatter is particularly important in pixelated detectors used for imaging as scatter escaping from an element reduces the energy deposited and absorbed scatter from a nearby element gives the appearance of additional low-energy incident photons. In this study, we describe a new approach for determining analytic models of absorbed energy distributions based on the established methods of linear cascaded-systems analysis (CSA) [12]–[20]. The CSA method accounts for spatial correlations and stochastic properties of energy-absorbing processes (photoelectric and Compton interactions) and has been successful for describing signal and noise performance and the detective quantum efficiency (DQE) of many X-ray detectors [20]. We apply it to a description of the AED of the five common convertor materials Si, Se, CsI, CdTe, and HgI , and validate the 0278-0062 © 2013 IEEE 1820 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 The probability that an incident X-ray photon with energy interacts in the X-ray convertor and deposits energy is given by the quantum absorption efficiency (2) Fig. 1. Schematic representation of the geometry used to calculate the reabsorption probabilities of characteristic and Compton-scatter photons for a semiinfinite slab with thickness . results with MC calculations. The method may be useful in an evaluation of other promising candidate compound convertor materials such as PbI , PbO, or TlBr. II. THEORY Fig. 1 illustrates an X-ray interaction in which a scatter photon is emitted. Energy is deposited at the primary interaction site and potentially at a remote reabsorption site. Scatter escape (from either front or rear detector surfaces) will result in a partial loss of incident energy and thereby dispersion of absorbed energy. In this model, all incident photons are treated identically and photoelectric and Compton interactions are each viewed as special cases of this simple interaction model. Rayleigh (coherent) scatter interactions do not deposit energy and are not included [20]. Similar to Chan and Doi [21], the X-ray convertor is subdivided into layers and solid-angle increments to derive analytic expressions for reabsorption probabilities. In all cases, incident photons are assumed normal to the X-ray convertor plane. Characteristic emissions and Compton scatter photons that are reabsorbed are assumed to deposit all of their energy with the exception of reabsorption of high-Z characteristic photons in low-Z atoms where another characteristic photon may be emitted. To keep the model simple, we consider only K-shell transitions and use the K photon energy averaged by the fraction of K-L and K-M transition similar to the approach used by Hajdok et al. [2]. A. X-ray Interaction Model Using the parallel cascades method of Yao and Cunningham [16], Fig. 2 illustrates the possible combinations of energy-depositing events in compound X-ray detector materials for a slab geometry. Each connecting arrow represents a spatial distribution of points that connect the output of one process to the input of the next and describe where in the image plane each process takes place. Each item connected by arrows represents a point process that acts on each point in its input distribution to produce a new distribution of points at its output. The initial input distribution represents the spatial distribution of X-ray photons incident on the detector. The final output distributions represent spatial distributions of energy-depositing events. is the sum of photoelectric and inwhere coherent linear interaction coefficients, respectively, and represents convertor-material thickness. The coherent (Rayleigh) scatter interaction coefficient is not included in since those interactions do not deposit energy or change the photon energy and do not affect the probability of photoelectric or incoherent scatter interactions. The probabilities of photoelectric and Compton scatter interactions are, therefore, given by (3) (4) 1) Photoelectric Interaction: In compound X-ray detector materials, PE absorption may occur in either the high or low-Z atom with probability given by (5) where Z denotes high or low-Z atoms when replaced with H and represent the fractional weight, or L, and density, and photoelectric attenuation coefficient, respectively. If the incident photon energy is greater than the binding energy of a K-shell electron, a PE interaction may result in the emission of a characteristic photon with probability where and represent the K-shell participation factor and fluorescence yield, respectively [22]. While there may be a characteristic photon emitted from other electron transitions, Hajdok et al. [2] showed that L-shell transitions are negligible. 2) Reabsorption of Characteristic Emission: Based on the technique described by Chan and Doi [21], we first consider the characteristic reabsorption probability for the semi-infinite slab geometry. The probability of generating a K photon in th layer is given by (6) where and represent the sublayer thickness and index, respectively. The probability of generating a K photon in an X-ray convertor with thickness is then given by (7) The characteristic photon will be emitted isotropically in the azimuthal angle . This symmetry allows us to use a quarter sphere (corresponding to and ) to calculate reabsorption properties. The set of solid angle elements corresponding to the first quadrant in YUN et al.: ANALYTIC MODEL OF ENERGY-ABSORPTION RESPONSE FUNCTIONS IN COMPOUND X-RAY DETECTOR MATERIALS 1821 Fig. 2. Schematic diagram describing the cascades of possible energy-depositing processes considered in this work including photoelectric emission escape or reabsorption, and Compton scatter escape or reabsorption. Each arrow represents a spatial distribution of points that connect the output of one process to the input of the next, describing where in the image plane each process takes place. Each item connected by arrows represents a process that acts on each point of the input distribution. For example, the first (left-most) triangle represents a quantum-selection process where each input point represents a photon that is either passed or not passed to the output according to the binomial probability . Each diamond represents a branch where each input point is passed to only one of the outputs according to the specified probabilities. AED is given by the combined probability of entries in each column (divided into photo peak, escape peak, and Compton continuum contributions for clarity). The shaded area describes reabsorption of a high-Z characteristic emission in a low-Z atom which results in an additional low-Z characteristic emission. azimuthal angle plane and is given by change in polar angle The probability that a characteristic photon generated in the th layer is reabsorbed in is given by This result gives the probability that an incident photon undergoes a photoelectric interaction and generates a characteristic X-ray that is reabsorbed in the convertor material. It is dependent on to reflect the energy dependence of interaction depth in the convertor. The probability of reabsorption per generated characteristic photon is then given by (9) (13) represents the Equation (13) gives the average probability that a characteristic photon is reabsorbed in the convertor material used to describe reabsorption in Fig. 2. 3) Compton Interaction: In a Compton-scatter event, energy is deposited at the primary interaction site and potentially at a remote reabsorption site and has recently been described in terms of cascaded-system analyses [20]. Unlike photoelectric emissions, the energy of Compton-scatter photons is a random variable determined by and scatter angle [22] (8) where represents the K-photon energy, and traveling path of the K photon (Fig. 1) (10) Summing (9) over all solid angles gives the reabsorption probability for a characteristic photon emitted from the th layer (11) (14) which is independent of the incident photon energy . The average K-photon reabsorption probability for a convertor with thickness is, therefore, given by (12) keV is the electron rest enwhere ergy and is a random variable with probability density function (PDF) given by [22]–[24] (15) 1822 where IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 represents the classical radius of the electron m and the Compton cross section per electron is given by [25] (16) The Klein–Nishina factor , is given by [22] (17) This expression for Klein–Nishina factor does not include the effects of bound electrons that might be important for high-Z atoms and low energies. The spectral density of Compton-scattered photons is then given by Fig. 3. Schematic representation of the geometry used to calculate the reabsorption probabilities of characteristic and Compton-scatter photons in a pixel with thickness and pixel pitch . The main difference between the pixel and slab geometries is the traveling path ( for the pixel and for the slab) of the characteristic/Compton-scatter photon. energy by is reabsorbed in the convertor material is then given (25) (18) 4) Reabsorption of Compton-Scatter Photons: The reabsorption probability of Compton-scatter photons is determined using a technique similar to above. The probability of a Compton event in the th layer of thickness is given by Summing over all layers gives the probability that the Comptonscatter photon is reabsorbed (26) (19) and the total probability of a Compton event is given by (20) The scatter spectrum is divided numerically into bins of energy and width keV with corresponding solid-angle increments of (21) where (22) (23) Subscripts lb and ub represent scatter angle lower and upper limits for a specified scatter-photon energy and the minimum scatter energy occurs when . The probability that a Compton-scatter photon with energy scattered in the th layer is emitted into solid angle is given by This result gives the probability that an incident photon interacts through the Compton effect and produces a Compton photon with energy that is reabsorbed in the convertor material. The probability of reabsorption per scattered photon is then given by (27) 5) Extension to Pixel Geometry: We extend the semi-infinite slab model to pixel geometries by defining square-shaped pixel boundaries in the plane, such as the regions of charge collection defined by electric fields in photoconductor-based direct-conversion detectors as illustrated in Fig. 3. For simplicity, we assume that incident photons arrive at the center of each element and ignore the effect of charge sharing (or optical scatter in a phosphor) between elements. While this may appear to be a limiting simplification, it is well recognized that photoncounting detectors must include some form of element “binning” to accurately determine the total deposited energy surrounding each interaction [26], [27], [5]. Our assumption therefore corresponds to the case of summing the signals from elements surrounding the primary interaction location. The horizontal distance in Fig. 3 is given by (24) where is given by (8), is the polar angle index which is a function of through and , and is given by (9). The probability that a scattered photon from the th layer with (28) where is the pixel pitch, in azimuthal angle and represents the change is the azimuthal angle YUN et al.: ANALYTIC MODEL OF ENERGY-ABSORPTION RESPONSE FUNCTIONS IN COMPOUND X-RAY DETECTOR MATERIALS index. The traveling path for the pixel geometry therefore, given by (Fig. 3) is, 1823 a step at . After differentiating, the probability density of depositing energy into the photopeak given a Compton event, , is given by (29) and represent the polar where angle index and change in polar angle, respectively [see (10)]. Note that is independent of the depth index and therefore only accounts for the element boundaries in the plane. By choosing the smaller of and our model accounts for pixel boundaries in all three dimensions. For example, if the incident photon interacts at the bottom of the X-ray convertor and the scattering angle is very low, then the photon will escape through the bottom surface and the traveling path is given by . Alternatively, if a secondary photon is emitted at an scattering angle close to then it may escape the pixel in the lateral direction and the traveling path is given by . The solid angle modified for the pixel geometry is given by (33) with area equal to the avresulting in a -function at erage reabsorption probability as a function of scatter energy and weighted by the scatter spectrum. The Compton continuum results when the scatter photon is not reabsorbed and only energy is deposited as represented by the entry in the Compton-continuum column. The integral probability is given by (34) (30) and differentiating with respect to energy gives the Compton continuum as The probability that a characteristic photon generated in the th layer is reabsorbed in is therefore given by (35) (31) The probability that a Compton photon generated in the th layer is reabsorbed in is given by 1) Slab Geometry: Combining photopeak results for photoelectric and Compton interactions gives the photopeak component of as (32) Summing over all , and gives the reabsorption probabilities of and Compton photons for the pixel geometry (similar to (13) and (27) for the slab geometry). B. Analytic Response Function is determined by The analytic response function considering all possible cascades in Fig. 2 resulting in energy deposition for any one interacting photon. For example, photoelectric interactions result in six cascade combinations that contribute to the photopeak and three to escape peaks, with each contribution corresponding to an entry in the columns labeled photopeak and escape peak in Fig. 2. The probability of each contribution is given by the product of all gains in the corresponding cascades and is equal to the combined probability density of depositing . To assist with these calculations, we define as the integral probability of depositing any energy between 0 and and therefore . Compton scatter is more complicated since the scatter-photon energy is represented by the random variable , resulting in a photopeak and Compton continuum. The photopeak occurs when the scatter photon is reabsorbed, as represented by the lowest entry in the photopeak column. The corresponding integral probability of depositing energy is zero for with (36) where . Similarly, escape peaks are given by (37) and the Compton continuum component by (38) Combining these results gives the energy response function (39) 2) Pixel Geometry: While the response function for a pixel geometry has a form similar to (39), reabsorption probabilities are lower due to shorter traveling paths which result in increased 1824 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 Fig. 4. Schematic illustration showing the effect of multiple scatter on the response function for slab (a) and pixel (b) geometries. Numbers in converter correspond to photon energy deposited with each interaction. Fig. 5. Schematic illustration of MC parameters used to calculate the AED. K-escape peak and Compton continuum, and absorption of characteristic and Compton-scatter photons escaping from neighboring pixels cause additional contributions. We represent these additional contributions as and let represent the resulting distribution of absorbed energies in pixel geometries as (40) where is given by (39) with the replacement of and with and . It is important to note that while describes the average distribution of energy deposited per photon incident on that element, includes absorption of characteristic emissions and Compton scatter from neighboring elements and therefore describes the average energy response from multiple photons incident on a detector array. By the principle of reciprocity, the average energy deposited is equal to the average energy that escapes laterally and is deposited in other elements, giving the spectral density of reabsorbed peaks (rp) as (41) and reabsorbed Compton scatter (rs) as (42) The total contribution is therefore . Fig. 4 illustrates the effect of multiple scatter on response functions. With the slab geometry, multiple scatter has little effect other than possibly altering the probability of scatter-photon escape. However, with pixel geometries, multiple scatter increases energy dispersion by moving energy from the photopeak to lower levels if a scatter photon leaves the element as well as adding contributions from neighboring elements. III. VALIDATION The analytic model is validated by MC simulations of X-ray absorption in Si, Se, CsI, CdTe, and HgI X-ray convertor materials. While Si is not a compound material, it has a high Compton and Rayleigh cross section with negligible characteristic photons over all diagnostic energy ranges and is therefore an important test of the Compton model. Se is also selected for validation as it has a single escape peak with negligible and energy differences. Both Se and CsI are widely used in digital radiography detectors and have important escape peaks. Material properties are taken from [22], [28], and [29]. A. Monte Carlo Simulation X-ray interaction and energy absorption was simulated using the particle tracking (pTrac) tally of the MC user code MCNP5 (Radiation Safety Information Computational Center (RSICC), Oak Ridge, TN, USA) for a uniformly distributed histories) as parallel-beam X-ray source (2 2 cm and illustrated in Fig. 5 for both slab and pixel geometries. The code was configured to generate data files describing all X-ray interactions for each photon history excluding electron transport considerations. These data files were used to determine the resulting distributions of energy-depositing events, scored in 1-keV bins, for both slab and pixel geometries. While both mono and poly-energetic (RQA5 [30]) X-ray spectra were used for slab geometries, only mono-energetic spectra were used for pixel geometries due to the need for very large storage requirements for each photon energy. The effects of multiple scatter on the response functions were determined by first including multiple scatter with each history and then repeating the analysis assuming all energy in the characteristic/scatter photons is completely deposited when reabsorbed. Also, MC simulations included Rayleigh scattering which was not included in the analytic models. Theoretical and MC calculations of the AED were compared in terms of the Swank factor, given by [4] (43) where where , and are the first three moments of the AED . The X-ray Swank factor takes YUN et al.: ANALYTIC MODEL OF ENERGY-ABSORPTION RESPONSE FUNCTIONS IN COMPOUND X-RAY DETECTOR MATERIALS Fig. 6. Comparison of analytic and MC response functions, with and without -thick multiple scattering, for the slab (a) and pixel (b) geometries using 500X-ray convertors and 100-keV incident photons. on values between zero and one with corresponding to an ideal detector, and is commonly used to describe the effects of energy dispersion on X-ray image quality [10], [31], [32], [3]. IV. RESULTS In the following results, sub-layer and angle indexes of and were used to minimize numerical errors which may be introduced by discretization of each dimension. A. Energy-Response Function 1) Semi-Infinite Slab Geometry: Fig. 6(a) shows a comparison of analytic and MC energy response functions, obtained both with and without multiple scattering, using 100 keV incident photons with 500- m-thick convertor materials. Qualitatively, there is very good agreement between the shape of the response functions in all cases with the differences being relatively small compared to other structures. All show the primary features of photo-peak, escape peaks and Compton continuum and differences are attributed to three considerations. The first is that analytic results show a simpler pattern of escape peaks in high-Z materials, particularly in HgI , due to the use of an average K-photon energy rather than separate and emission lines. The second is due to neglecting the effect of orbital (bound) electrons in Compton scattering in the analytic model which misses some energy broadening around the Compton edge [33]–[35]. The third is the effect of ignoring multiple scatter in the analytic model which is more difficult to see 1825 in Fig. 6 and is identified by comparing the MC results obtained with and without multiple scatter enabled in the calculation. Multiple scatter causes a small reduction in the energy of characteristic emissions and scatter photons that leave the detector, resulting in a small decrease in the intensity of escape peaks and the Compton continuum with a corresponding increase at energies just above the escape peaks and continuum. The close agreement with MC results gives confidence that Rayleigh scattering, ignored in the analytic model, has little influence on the response functions. 2) Pixel Geometry: Fig. 6(b) illustrates energy-response functions for the pixelated geometry for 100 keV incident photon energy, 500- m-thick X-ray convertors and a pixel pitch of 100 m. In general, the distribution of absorbed energies for the pixel geometry is more dispersed than for the slab geometry due to new contributions from absorption of characteristic and Compton photons from neighboring elements. In addition, escape peaks are larger due to the increased probability of escape from small detector elements. In particular, absorption of Compton photons generated in neighboring elements results in a continuous distribution of absorbed energies for all results at energies just below the photopeaks. However, the analytic calculations do not predict reabsorption of fluorescence and Compton photons from multiple scatter in the 30–70 keV energy range, as seen in Fig. 6(b). In general, scatter has a greater effect on the response function for pixel geometries than slab geometries, as shown in Fig. 4. 3) Energy Dependency: Fig. 7 shows a comparison of the response function for slab geometry expressed as a function of incident photon energy from analytic and MC calculations. The color scales show the display range of each paired comparison. In all cases, there is good agreement between the analytic model and MC results with minor discrepancies in escape peaks for the same reasons discussed above. The Compton continuum is not visible in all but Si results due to the very low response value compared to the photopeak height. B. Absorbed Energy Distribution Fig. 8 shows AED curves obtained using analytic and MC calculations including multiple scatter for an RQA-5 X-ray spectrum. The solid and dotted lines correspond to analytic and MC results respectively for the slab geometry. Overall, there is excellent agreement between these two methods. Escape peaks are not obvious due to the broad X-ray spectrum. Similar to energy response results above, MC calculations show a broadening of the Compton edge although this is visible in only the Si and Se results where Compton plays a substantial role. The effect of multiple scatter depends on escape peak energies and is less obvious for the broad spectrum, but generally causes a reduction in AED values in the 15–35 keV range, depending on escape-peak energies. Very little effect is seen for HgI since the K-edge energy of Hg is above the RQA-5 spectrum maximum energy (70 kV). Analytic results for the pixel geometry are represented by the dashed lines in Fig. 8. As expected, the higher escape probability of emission/scatter photons results in more low-energy absorption events, and therefore an increase in energy dispersion. In 1826 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 Fig. 7. Comparison of the 2-D response function calculated by analytic (a) and MC (b) methods for incident energies ranging from 1 to 100 keV for a 500-thick semi-infinite slab. Fig. 8. Comparison of analytic and MC AEDs for 500- m-thick X-ray convertors with an RQA5 X-ray spectrum. The solid, dotted, and dashed lines represent calculations for the analytic semi-infinite slab, MC semi-infinite slab (including multiple scatter), and analytic pixel geometry, respectively. addition, reabsorption peaks from characteristic photons generated in neighboring elements appear in the AEDs because they are independent of incident photon energy. Table I summarizes the X-ray Swank factor values corresponding to AED curves in Fig. 8. In all cases, the analytic and MC results for the slab geometry agree within 3% and show similar trends. Si has the lowest value due to the greatest Compton continuum relative to the photopeak. For all other materials, MC results are slightly greater than analytic because of less dispersion due to multiple scatter. It is interesting to note that for this spectrum, the largest Swank factors for the slab geometry occur with Se and HgI and are 6%–7% greater than CsI. However, this difference is smaller with the pixel geometry. C. Effects of Convertor Thickness and Pixel Size Fig. 9(a) and (b) illustrates the energy-response functions for CdTe with 100-keV incident photons and selected X-ray con- TABLE I X-RAY SWANK NOISE FACTORS DETERMINED FROM AED CURVES IN FIG. 8 FOR RQA-5 SPECTRA vertor thicknesses for slab and pixel geometries. Increased convertor thickness results in increased photo-peak, escape peaks, and Compton continuum due to the greater quantum absorption efficiency for both slab and pixel geometry. Fig. 9(c) shows the effect of pixel size for CdTe with 100-keV incident photons. Larger pixel sizes result in greater photopeaks due to the longer path length of emission/scatter photons in the YUN et al.: ANALYTIC MODEL OF ENERGY-ABSORPTION RESPONSE FUNCTIONS IN COMPOUND X-RAY DETECTOR MATERIALS 1827 Fig. 9. Comparisons of analytic and MC response functions for CdTe using slab (a) and pixel [(b) and (c)] geometries for selected convertor thicknesses and pixel pitches for 100 keV incident photons. primary pixel volume. The effect of photon escape from neighboring elements in the pixel geometry increases as the element size is reduced, resulting in increased energy dispersion. V. DISCUSSION AND CONCLUSION We have presented an analytic technique for obtaining the distribution of absorbed X-ray energies in compound X-ray detector materials for both semi-infinite slab and pixel geometries. The pixel geometry will be particularly important in describing the imaging performance of pixelated detectors used in conventional X-ray imaging and novel photon-counting applications. The results from our analytic approach agreed well with results from MC calculations. The strength of the analytic approach is that it allows us to obtain the 2-D response function of arbitrary compound materials without time-consuming MC simulations. It does not include the effects of multiple scattering or outer-shell (L or M) transitions although MC calculations show these to be relatively small in the cases tested. Also, the analytic model does not describe charge or optical photon transport phenomena (e.g., charge sharing in a photoconductor or spreading of optical photons in a phosphor) and electrical properties (e.g., mobility and work energy) of X-ray detectors. While the analytic model provides some physical insight into detector performance, particularly in the design of photon-counting applications and detector optimization, it must also be considered an approximation that may produce optimistic results. REFERENCES [1] H. K. Kim, C. H. Lim, J. Tanguay, S. Yun, and I. A. Cunningham, “Spectral analysis of fundamental signal and noise performances in photoconductors for mammography,” Med. Phys., vol. 39, no. 5, pp. 2478–2490, 2012. [2] G. Hajdok, J. Yao, J. J. Battista, and I. A. Cunningham, “Signal and noise transfer properties of photoelectric interactions in diagnostic X-ray imaging detectors,” Med. Phys., vol. 33, no. 10, pp. 3601–3620, 2006. [3] J. Tanguay, H. K. Kim, and I. A. Cunningham, “The role of X-ray Swank factor in energy-resolving photon-counting imaging,” Med. Phys., vol. 37, no. 12, pp. 6205–6211, 2010. [4] R. K. Swank, “Absorption and noise in X-ray phosphors,” J. Appl. Phys., vol. 44, no. 9, pp. 4199–4203, 1973. [5] H. Ding and S. Molloi, “Image-based spectral distortion correction for photon-counting X-ray detectors,” Med. Phys., vol. 39, no. 4, pp. 1864–1876, 2012. [6] J. P. Schlomka, E. Roessl, R. Dorscheid, S. Dill, G. Martens, T. Istel, C. Bäumer, C. Herrmann, R. Steadman, G. Zeitler, A. Livne, and R. Proksa, “Experimental feasibility of multi-energy photon-counting K-edge imaging in pre-clinical computed tomography,” Phys. Med. Biol., vol. 53, no. 15, pp. 4031–4047, 2008. [7] D. A. Jaffray, J. J. Battista, A. Fenster, and P. Munro, “Monte Carlo studies of X-ray energy absorption and quantum noise in megavoltage transmission radiography,” Med. Phys., vol. 22, no. 7, pp. 1077–1088, 1995. [8] K. Taguchi, M. Zhang, E. C. Frey, X. Wang, J. S. Iwanczyk, E. Nygard, N. E. Hartsough, B. M. W. Tsui, and W. C. Barber, “Modeling the performance of a photon counting X-ray detector for CT: Energy response and pulse pileup effects,” Med. Phys., vol. 38, no. 2, pp. 1089–1102, 2011. [9] Y. Fang, A. Badal, N. Allec, K. S. Karim, and A. Badano, “Spatiotemporal Monte Carlo transport methods in X-ray semiconductor detectors: Application to pulse-height spectroscopy in a-Se,” Med. Phys., vol. 39, no. 1, pp. 308–319, 2012. 1828 [10] I. M. Blevis, D. C. Hunt, and J. A. Rowlands, “X-ray imaging using amorphous selenium: Determination of Swank factor by pulse height spectroscopy,” Med. Phys., vol. 25, no. 5, pp. 638–641, 1998. [11] R. J. LeClair, Y. Wang, P. Zhao, M. Boileau, L. Wang, and F. Fleurot, “An analytic model for the response of a CZT detector in diagnostic energy dispersive X-ray spectroscopy,” Med. Phys., vol. 33, no. 5, pp. 1329–1337, 2006. [12] M. Rabbani, R. Shaw, and R. Van Metter, “Detective quantum efficiency of imaging systems with amplifying and scattering mechanisms,” J. Opt. Soc. Am. A, vol. 4, no. 5, pp. 895–901, 1987. [13] I. A. Cunningham, M. S. Westmore, and A. Fenster, “A spatial-frequency dependent quantum accounting diagram and detective quantum efficiency model of signal and noise propagation in cascaded imaging systems,” Med. Phys., vol. 21, no. 3, pp. 417–427, 1994. [14] J. H. Siewerdsen, L. E. Antonuk, Y. El-Mohri, J. Yorkston, W. Huang, and I. A. Cunningham, “Signal, noise power spectrum, and detective quantum efficiency of indirect-detection flat-panel imagers for diagnostic radiology,” Med. Phys., vol. 25, no. 5, pp. 614–628, 1998. [15] I. A. Cunningham and R. Shaw, “Signal-to-noise optimization of medical imaging systems,” J. Opt. Soc. Am. A, vol. 16, no. 3, pp. 621–632, Mar. 1999. [16] J. Yao and I. A. Cunningham, “Parallel cascades: New ways to describe noise transfer in medical imaging systems,” Med. Phys., vol. 28, no. 10, pp. 2020–2038, 2001. [17] A. Ganguly, S. Rudin, D. R. Bednarek, and K. R. Hoffmann, “Microangiography for neuro-vascular imaging. II. Cascade model analysis,” Med. Phys., vol. 30, no. 11, pp. 3029–3039, 2003. [18] S. Vedantham, A. Karellas, and S. Suryanarayanan, “Solid-state fluoroscopic imager for high-resolution angiography: Parallel-cascaded linear systems analysis,” Med. Phys., vol. 31, no. 5, pp. 1258–1268, 2004. [19] S. Yun, C. H. Lim, H. K. Kim, J. Tanguay, and I. A. Cunningham, “Finding the best photoconductor for digital mammography detectors,” Nucl. Instr. Meth. Phys. Res. A, vol. 652, no. 1, pp. 829–833, 2011. [20] S. Yun, J. Tanguay, H. K. Kim, and I. A. Cunningham, “Cascaded-systems analysis and the detective quantum efficiency of single-Z X-ray detectors including photoelectric, coherent and incoherent interactions,” Med. Phys., vol. 40, no. 4, p. 041916, 2013. [21] H. P. Chan and K. Doi, “Energy and angular dependence of X-ray absorption and its effect on radiographic response in screen-film systems,” Phys. Med. Biol., vol. 28, no. 5, pp. 565–579, 1983. [22] F. H. Attix, Introduction to Radiological Physics and Radiation Dosimetry. New York: Wiley, 1986. IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 32, NO. 10, OCTOBER 2013 [23] J. Yao and I. A. Cunningham, “Compton scatter in frequency space: A theoretical study,” in Proc. SPIE, 2002, vol. 4682, no. 1, pp. 479–490. [24] G. Hajdok, J. J. Battista, and I. A. Cunningham, “Fundamental X-ray interaction limits in diagnostic imaging detectors: Spatial resolution,” Med. Phys., vol. 35, no. 7, pp. 3180–3193, 2008. [25] C. M. Davisson and R. D. Evans, “Gamma-ray absorption coefficients,” Rev. Mod. Phys., vol. 24, pp. 79–107, Apr. 1952. [26] R. Ballabriga, M. Campbell, E. H. M. Heijne, X. Llopart, and L. Tlustos, “The medipix3 prototype, a pixel readout chip working in single photon counting mode with improved spectrometric performance,” IEEE Trans. Nucl. Sci., vol. 54, no. 5, pp. 1824–1829, Oct. 2007. [27] C. Ponchut, “Correction of the charge sharing in photon-counting pixel detector data,” Nucl. Instrum. Meth. Phys. Res. A, vol. 591, no. 1, pp. 311–313, 2008. [28] S. T. Perkins, M. H. Chen, D. E. Cullen, and J. H. Hubbell, Tables and graphs of atomic subshell and relaxation data derived from the LLNL evaluated Atomic Data Library (EADL), Z=1-100 Lawrence Livermore Nat. Lab., UCRL-50400, 1991, vol. 30. [29] C. Koughia, S. Kasap, and P. Capper, Springer Handbook of Electronic and Photonic Materials. New York: Springer, 2006. [30] Medical electrical equipment-characteristics of digital X-ray imaging devices—Part 1: Determination of the detective quantum efficiency Int. Electrotechn. Commission, Med. Electr. Equip., IEC 62220-1, 2003. [31] W. Zhao, W. G. Ji, and J. A. Rowlands, “Effects of characteristic x rays on the noise power spectra and detective quantum efficiency of photoconductive X-ray detectors,” Med. Phys., vol. 28, no. 10, pp. 2039–2049, 2001. [32] G. Hajdok, J. J. Battista, and I. A. Cunningham, “Fundamental X-ray interaction limits in diagnostic imaging detectors: Frequency-dependent Swank noise,” Med. Phys., vol. 35, no. 7, pp. 3194–3204, 2008. [33] A. Sood and M. C. White, Doppler energy broadening for incoherent scattering in MCNP5, Part II, LA-UR-04-0488 LANL, Tech. Rep., 2004. [34] C. Z. Uche, M. J. Cree, and W. H. Round, “GEANT4 simulation of the effects of Doppler energy broadening in Compton imaging,” Australas. Phys. Eng. Sci. Med., vol. 34, no. 3, pp. 409–414, 2011. [35] L. J. Bartol and L. A. DeWerd, “Technical note: Improved implementation of doppler broadening in MCNP5,” Med. Phys., vol. 39, no. 9, pp. 5635–5638, 2012.
© Copyright 2026 Paperzz