Search

• Proposed a total-variation constrained iterative iso -transmission line material decomposition algorithm. • Suppress noise amplification while performing basis material decomposition. • Provide superior quantitative imaging accuracy with reduced noise compared with traditional material decomposition methods. Photon-counting computed tomography (PCCT) stands out for its remarkable material distinguishability, presenting a significant advancement over traditional CT. However, noise amplification occurs in decomposed basis material images, attributed to the overlap of energy spectra and the correlation among basis functions. To suppress noise while improving the accuracy of quantitative imaging in material decomposition, this study aims to propose a total-variation constrained iterative iso -transmission line (TVITL) material decomposition algorithm for multi-energy PCCT. The TVITL algorithm transforms the numerical solution of the pixel-by-pixel equations into a matrix operation to holistically solve the basis material decomposition and thus incorporates an image noise penalty weight and a full-variation regularization term for denoising. By effectively integrating direct smoothing information from neighbouring pixels, the algorithm accurately accounts for the noise characteristics, thereby improving the accuracy of quantitative imaging during the material decomposition process. The proposed method was evaluated by comparing it with the traditional non-linear equations (NLE), iso -transmission line (ITL), and A-table methods. We compared the four material decomposition methods in calibration reliability tests and quantitative imaging experiments. The proposed algorithm has higher thickness estimation and quantitative imaging accuracy and the lowest noise standard deviation than the other three algorithms, whether it is a highly noisy simulation experiment with ideal detector response or a real system with distortion of detector response spectrum. Our method can perform spectral CT material decomposition well under simulation and experiment situations. Notably, it excels in noise suppression and exhibits commendable accuracy in material discrimination. [ABSTRACT FROM AUTHOR]

In this paper, an image up-sampling method for a dual-radiation nondestructive cargo inspection system, that uses x-ray and neutron, is proposed. The inspection system is assumed to have a larger neutron detector pixel size than that of x-ray detector pixel, which is usually the case to address the lower neutron source flux and the lower detector efficiency than those in the x-ray imaging system. The proposed technique adopts a cluster map of an x-ray projection image with high spatial resolution to up-sample the corresponding neutron projection image with low spatial resolution. The cluster map is acquired by use of a modified k-means clustering algorithm. The up-sampling process is formulated by a linear least-square problem of mapping the neutron projection image onto the cluster map. Consequently, the neutron projection image would have the same dimension with the x-ray projection image array allowing R-value based material decomposition. We have conducted a numerical study using GATE simulation to demonstrate the efficacy of the proposed method. The proposed method was compared with a linear interpolation method and also with an up-sampling technique that does not use the cluster map. The outperformance of the proposed method over those reference techniques has been successfully presented. [ABSTRACT FROM AUTHOR]

Rationale and Objectives: To evaluate the material decomposition (MD) techniques in rapid kVp switching dual-energy CT (rsDECT) for quantifying liver fat content in rats with nonalcoholic fatty liver.Materials and Methods: Fifty male Sprague-Dawley (SD) rats were divided into study group (n=37) and control group (n = 13) and underwent rsDECT examination at different intervals. All the data analysis was performed using AW4.7 workstation. The fat contents under the traditional fat(water), fat(blood), and fat(muscle) material decomposition (MD) images and the fat volume fraction (FVF) from the liver fat maps generated using multi-material decomposition (MMD) technique were measured. The pathological grades (grade 0, 1, 2 and 3) of fatty liver were determined after euthanasia. The measurement differences among different grades and the correlation of measurements with different grades was analyzed using ANOVA and Spearman correlation, respectively. A receiver operating characteristics (ROC) curve was used to analyze the diagnostic efficacies of fat contents and FVF.Results: There were statistically significant differences in FVF and fat contents under fat(water), fat(blood), fat(muscle) based MD images among different grades. These values correlated well with the pathological grades (R-value: 0.90, 0.75, 0.79, 0.80, all p <0.001), with FVF having the highest correlation. The area-under-the-curve in ROC of using FVF was the highest, with the cut-off value of 0.92 for sensitivity of 89.2% and specificity of 100%.Conclusion: The rsDECT MD techniques could quantitatively evaluate the fat content of fatty liver in rat, with the FVF from MMD having the highest correlation with pathological grades. [ABSTRACT FROM AUTHOR]

Dual-energy CT (DECT) can improve the accuracy of myocardial perfusion CT with projection-based monochromatic (DECT-MCE) and quantification of myocardial iodine in material decomposition (DECT-MD) reconstructions. However, evaluation of multiple reconstructions is laborious and the optimal reconstruction to detect myocardial perfusion defects is unknown. Left ventricular (LV) phantoms with artificial perfusion defects were scanned using DECT and single energy cardiac computed tomography angiography (SECT). Reconstructions of DECT-MCE at 40, 70, 100 and 140 keV, DECT-MD pairs of water, iodine, iron and fat, and SECT were evaluated using a 17-segment myocardial model. The diagnostic performance of each reconstruction was calculated on a per-segment basis and compared across DECT reconstructions. Over 34 phantoms with artificial perfusion defects were found in 64/578 (11%) of segments, the sensitivity of DECT-MCE at 40, 70, 100, and 140 keV was 100% (95% confidence interval (CI): 93–100), 100% (95% CI: 93–100), 71% (95% CI: 56–83), and 25% (95% CI: 14–40), respectively, with a significant decline between 70 keV and 100 keV (p < 0.001). The specificity of DECT-MCE was 100% at all energies (95% CI: 99–100). As a group, the DECT-MD iodine background reconstructions had significantly lower sensitivity than the remaining modes (2.1% [95% CI, 0.05–11.1], vs. 100% [95% CI, 92.6–100], p < 0.001). Specificity of all material pair modes remained 100%. Using LV phantom models, the approach with the best sensitivity and specificity to assess myocardial perfusion defects with DECT are reconstructions of DECT-MCE at 40 or 70 KeV and DECT-MD without iodine background. • Multiple reconstructions can evaluate myocardial perfusion with projection-based dual-energy CT (DECT) • Myocardial phantoms were scanned with monochromatic (DECT-MCE) and material decomposition (DECT-MD) reconstructions • DECT-MCE 40 and 70 keV are more accurate than higher keV. DECT-MD pairs with iodine background are less accurate than others. • The most efficient diagnostic strategy is to use DECT-MCE at 40 or 70 keV, or DECT-MD without iodine background [ABSTRACT FROM AUTHOR]

Rationale and Objectives: Iodine quantification (IQ) and virtual noncontrast (VNC) images produced by dual-energy CT (DECT) can be used for various clinical applications. We investigate the performance of dual-layer DECT (DLDECT) in different phantom sizes and varying radiation doses and tube voltages, including a low-dose pediatric setting.Materials and Methods: Three phantom sizes (simulating a 10-year-old child, an average, and a large-sized adult) were scanned with iodine solution inserts with concentrations ranging 0-32 mg/ml, using the DLDECT. Each phantom size was scanned with CTDIvol 2-15 mGy at 120 and 140 kVp. The smallest phantom underwent additional scans with CTDIvol 0.9-1.8 mGy. All scans were repeated 3 times. Each iodine insert was analyzed using VNC and IQ images for accuracy and precision, by comparison to known values.Results: For scans from 2 to 15 mGy mean VNC attenuation and IQ error in the iodine inserts in the small, medium, and large phantoms was 1.2 HU ± 3.2, -1.2 HU ± 14.9, 2.6 HU ± 23.6; and +0.1 mg/cc ± 0.4, -0.9 mg/cc ± 0.9, and -1.8 mg/cc ± 1.8, respectively. In this dose range, there were no significant differences (p ≥ 0.05) in mean VNC attenuation or IQ accuracy in each phantom size, while IQ was significantly less precise in the small phantom at 2 mGy and 10 mGy (p < 0.05). Scans with CTDIvol 0.9-1.8 mGy in the small phantom showed a limited, but statistically significantly lower VNC attenuation precision and IQ accuracy (-0.5 HU ± 5.3 and -0.3 mg/cc ± 0.5, respectively) compared to higher dose scans in the same phantom size.Conclusion: Performance of iodine quantification and subtraction by VNC images in DLDECT is largely dose independent, with the primary factor being patient size. Low-dose pediatric scan protocols have a significant, but limited impact on IQ and VNC attenuation values. [ABSTRACT FROM AUTHOR]

Abstract Surface damage and material failure mechanism of CaF 2 single crystal by 355 nm/6.8 ns laser pulses are systematically studied in this work. Surface damage characteristics, such as damage threshold, damage growth threshold, damage growth laws and damage growth coefficient of CaF 2 crystal are carefully studied. Surface material cracking and local material vitrification (from CaF 2 crystal to defective CaF 2-x materials) are revealed. Material decomposition, oxidation and formation of Ca0, Ca2+and (F−-F−) centers in surface layer are illustrated. Laser-driven thermal accumulation and thermal-stress gradient in CaF 2 crystal are calculated and discussed based on the theory of Fourier heat conduction and thermal-elastic constitutive equations. Theory analysis suggests that the material failure process in CaF 2 crystal is mainly driven by significant thermal stress gradient. Thermal gradient induce the material decomposition and local vitrification in CaF 2 crystal. Combined results on surface damage and material failure behavior are beneficial to understand nanosecond laser-driven modifications and damage in fluoride crystals at ultraviolet wavelength. Highlights • Laser-driven surface damage characteristics and surface modifications were studied. • Local material vitrification and decomposition of CaF 2 surface were revealed. • Laser-driven surface cracking and significant hoop stress of CaF 2 were disclosed. [ABSTRACT FROM AUTHOR]

Purpose of this study is to re-evaluate the accuracy and diagnostic reliability of virtual non-contrast (VNC) images acquired with the photon-counting computed tomography (PCCT) after an update of the CT scanner software. Fifty-four patients were retrospectively enrolled. VNC images were reconstructed from true non-contrast (TNC) images (VNCn) and contrast-enhanced images in portal venous contrast phase (VNCv). Additionally, a liver-specific VNC (VNCl) was assessed. Quantitative image properties of VNC and TNC images were compared and consistency between VNC images was evaluated. Regions of interest were drawn in the liver, spleen, renal cortex, aorta, muscle and subcutaneous fat. Attenuation values on all VNC images differed significantly from TNC images in the liver, renal cortex, aorta and fat. A mean offset of <10HU between TNC and all VNC images was found in the liver, spleen and muscle. The comparison of TNC and VNCl images revealed an offset < 10HU in fat. Differences ≤ 10HU between TNC and VNCv and between TNC and VNCl were found in 68%, respectively in 75%. Differences ≤ 15HU were found in 79%, respectively in 92% of all measurements. Differences ≤ 10HU between TNC and VNCn were found in 79% and differences ≤ 15HU in 85%. Although there are statistically significant differences between HU values measured on TNC and VNC images in certain tissues, the minor offsets measured in liver and spleen suggest a good clinical applicability of VNCv and VNCl images. The significantly lower offset in subcutaneous fat on VNCl images suggests a superiority for measurements in adipose tissues. [ABSTRACT FROM AUTHOR]

A decomposition model with an iterative solving framework is proposed for use in MeV dual energy computed tomography (DECT). The decomposition model can deal with a broad range of materials and better matches the physics of MeV photons. The proposed iterative solving framework is based on the conventional projection-domain decomposition method, and verified by numerical experiments and experiments. It is shown that images reconstructed with the proposed method have less artifacts and provide more accurate estimates of the atomic number of materials than the conventional projection-domain decomposition method. [ABSTRACT FROM AUTHOR]

Computed tomography (CT) has a major advantage over radiography in that it provides 3D imaging and resolves the structural overlap problem inherent to radiography. Since the early 1970s, CT evolved in parallel with radiography in many areas. As megavoltage (MV) cargo radiography is now in use, it is of interest to evaluate the feasibility of corresponding MV cargo CT. A feasibility study of MV cargo CT has been performed in this work. The MV cargo CT system has an imaging field of view of 3.25 m in diameter. The system design includes a compact X-ray source and detectors that are already used in existing MV cargo radiography systems. Although compact X-ray sources provide limited X-ray output, it is shown that this output is sufficient to perform CT imaging of large cluttered cargo contents for which MV radiography may be inconclusive. The scan time is 60 s at the X-ray dose levels lower than the permitted limit of 5 mGy. A steel tank phantom with 2 m diameter including different materials was used. The MV CT images were generated at 3.5 MV, 6 MV, and 9 MV beam energies. The materials were separated based on their CT numbers as well as using dual energy subtraction method. The materials with close densities and X-ray attenuations such as W and U were reliably separated and quantified based on their measured CT numbers. The boundary conditions for the system design were determined, including scan time, radiation dose, maximum object thicknesses, and spatial resolution. Based on the findings of this feasibility study, it is concluded that the MV cargo CT system with large field of view can be developed based on existing technology already used in MV cargo radiography. Such a CT system can be extremely useful when used adjunct to existing MV radiography on resolving the special cases that cannot be resolved conclusively with radiography alone. [ABSTRACT FROM AUTHOR]

Purpose: To evaluate virtual-non-contrast (VNC) images obtained from clinical triphasic scans with a dual-layer spectral computed tomography system regarding accuracy of iodine subtraction.Material and Methods: From September to December 2016, 62 consecutive patients who underwent a clinical routine triphasic CT examination were included into this retrospective study. VNC images based on the arterial and portal venous phase were generated. For every patient and every contrast phase, a region-of-interest (ROI) was defined in aorta, liver, renal cortex, spongious bone, fat, muscle and fluid (i.e. gallbladder, urinary bladder), resulting in 2170 ROIs. VNC images were compared to true-non-contrast (TNC) images regarding difference in attenuation. Consistency between VNC images obtained from the arterial and portal venous phase as well as the influence of the initial attenuation on respective VNC images were evaluated.Results: Comparison of HU in VNC and TNC images showed a high accuracy of iodine elimination. Mean difference between TNC and VNC images was only 0.5 ± 8.5 HU and >90% of all comparisons showed a difference of less than 15 HU. For all tissues but spongious bone, mean absolute difference between TNC and VNC images was below 10 HU. VNC images derived from the arterial and the portal venous phase showed excellent correlation. The quality of iodine removal in VNC images was not influenced by the original contrast enhancement. However, VNC images cannot be used for evaluation of iodine removal in bone as bone and iodine can hardly be differentiated via spectral CT.Conclusion: VNC imaging in DL-CT is a promising tool for daily clinical routine. As non-enhanced CT images are essential in multiple clinical situations, the permanent availability of VNC images with dual-layer spectral CT will result in a substantial reduction of radiation exposure and an increased diagnostic value of monophasic contrast-enhanced CT scans. [ABSTRACT FROM AUTHOR]

Iron is an essential element in the human body and disorders in iron such as iron deficiency or overload can cause serious diseases. This paper aims to explore the ability of spectral X-ray CT to quantitatively separate iron from calcium and potassium and to investigate the influence of different acquisition parameters on material decomposition performance. We simulated spectral X-ray CT imaging of a PMMA phantom filled with iron, calcium, and potassium solutions at various concentrations (15–200 mg/cc). Different acquisition parameters were considered, such as the number of energy bins (6, 10, 15, 20, 30, 60) and exposure factor per projection (0.025, 0.1, 1, 10, 100 mA s). Based on the simulation data, we investigated the performance of two regularized material decomposition approaches: projection domain method and image domain method. It was found that the former method discriminated iron from calcium, potassium and water in all cases and tended to benefit from lower number of energy bins for lower exposure factor acquisition. The latter method succeeded in iron determination only when the number of energy bins equals 60, and in this case, the contrast-to-noise ratios of the decomposed iron images are higher than those obtained using the projection domain method. The results demonstrate that both methods are able to discriminate and quantify iron from calcium, potassium and water under certain conditions. Their performances vary with the acquisition parameters of spectral CT. One can use one method or the other to benefit better performance according to the data available. [ABSTRACT FROM AUTHOR]

Purpose: Evaluation of the influence of iterative reconstruction, tube settings and patient habitus on the accuracy of iodine quantification with dual-layer spectral CT (DL-CT).Material and Methods: A CT abdomen phantom with different extension rings and four iodine inserts (1, 2, 5 and 10 mg/ml) was scanned on a DL-CT. The phantom was scanned with tube-voltages of 120 and 140 kVp and CTDIvol of 2.5, 5, 10 and 20 mGy. Reconstructions were performed for eight levels of iterative reconstruction (i0-i7). Diagnostic dose levels are classified depending on patient-size and radiation dose.Results: Measurements of iodine concentration showed accurate and reliable results. Taking all CTDIvol-levels into account, the mean absolute percentage difference (MAPD) showed less accuracy for low CTDIvol-levels (2.5 mGy: 34.72%) than for high CTDIvol-levels (20 mGy: 5.89%). At diagnostic dose levels, accurate quantification of iodine was possible (MAPD 3.38%). Level of iterative reconstruction did not significantly influence iodine measurements. Iodine quantification worked more accurately at a tube voltage of 140 kVp. Phantom size had a considerable effect only at low-dose-levels; at diagnostic dose levels the effect of phantom size decreased (MAPD <5% for all phantom sizes).Conclusion: With DL-CT, even low iodine concentrations can be accurately quantified. Accuracies are higher when diagnostic radiation doses are employed. [ABSTRACT FROM AUTHOR]

Dual-energy X-ray inspection systems are widely used today for it provides X-ray attenuation contrast of the imaged object and also its material information. Material decomposition capability allows a higher detection sensitivity of potential targets including purposely loaded impurities in agricultural product inspections and threats in security scans for example. Dual-energy X-ray transmission data can be transformed into two basis material thickness data, and its transformation accuracy heavily relies on a calibration of material decomposition process. The calibration process in general can be laborious and time consuming. Moreover, a conventional calibration method is often challenged by the nonuniform spectral characteristics of the X-ray beam in the entire field-of-view (FOV). In this work, we developed an efficient material decomposition calibration process for a linear accelerator (LINAC) based high-energy X-ray cargo inspection system. We also proposed a multi-spot calibration method to improve the decomposition performance throughout the entire FOV. Experimental validation of the proposed method has been demonstrated by use of a cargo inspection system that supports 6 MV and 9 MV dual-energy imaging. [ABSTRACT FROM AUTHOR]

Megavoltage (MV) radiography has important applications in imaging large cargos for detecting illicit materials. A useful feature of MV radiography is the possibility of decomposing and quantifying materials with different atomic numbers. This can be achieved by imaging cargo at two different X-ray energies, or dual energy (DE) radiography. The performance of both single energy and DE radiography depends on beam energy, beam filtration, radiation dose, object size, and object content. The purpose of this work was to perform comprehensive qualitative and quantitative investigations of the image quality in MV radiography depending on the above parameters. A digital phantom was designed including Fe background with thicknesses of 2cm, 6cm, and 18cm, and materials samples of Polyethylene, Fe, Pb, and U. The single energy images were generated at x-ray beam energies 3.5MV, 6MV, and 9MV. The DE material decomposed images were generated using interlaced low and high energy beams 3.5/6MV and 6/9MV. The X-ray beams were filtered by low-Z (Polyethylene) and high-Z (Pb) filters with variable thicknesses. The radiation output of the accelerator was kept constant for all beam energies. The image quality metrics was signal-to-noise ratio (SNR) of the particular sample over a particular background. It was found that the SNR depends on the above parameters in a complex way, but can be optimized by selecting a particular set of parameters. For some imaging setups increased filter thicknesses, while strongly absorbing the beams, increased the SNR of material decomposed images. Beam hardening due to polyenergetic x-ray spectra resulted in material decomposition errors, but this could be addressed using region of interest decomposition. It was shown that it is not feasible to separate the materials with close atomic numbers using the DE method. Particularly, Pb and U were difficult to decompose, at least at the dose levels allowed by radiation source and safety requirements. [ABSTRACT FROM AUTHOR]

• A virtual CBCT simulator with 36 head and neck human phantoms was developed. • Virtual CBCT images were generated by direct and scattered X-ray on a flat panel detector. • Based on the virtual CBCT images, an elemental material decomposition (EMD) was developed using Deep Learning-based EMD model. • Successful prediction of EMD for real patients/phantoms were given. • This study showed a potential of using computer vision for medical data preparation and analysis in adaptive radiation therapy. The purpose of this study is to develop a virtual CBCT simulator with a head and neck (HN) human phantom library and to demonstrate the feasibility of elemental material decomposition (EMD) for quantitative CBCT imaging using this virtual simulator. The library of 36 HN human phantoms were developed by extending the ICRP 110 adult phantoms based on human age, height, and weight statistics. To create the CBCT database for the library, a virtual CBCT simulator that simulated the direct and scattered X-ray on a flat panel detector using ray-tracing and deep-learning (DL) models was used. Gaussian distributed noise was also included on the flat panel detector, which was evaluated using a real CBCT system. The usefulness of the virtual CBCT system was demonstrated through the application of the developed DL-based EMD model for case involving virtual phantom and real patient. The virtual simulator could generate various virtual CBCT images based on the human phantom library, and the prediction of the EMD could be successfully performed by preparing the CBCT database from the proposed virtual system, even for a real patient. The CBCT image degradation owing to the scattered X-ray and the statistical noise affected the prediction accuracy, although these effects were minimal. Furthermore, the elemental distribution using the real CBCT image was also predictable. This study demonstrated the potential of using computer vision for medical data preparation and analysis, which could have important implications for improving patient outcomes, especially in adaptive radiation therapy. [ABSTRACT FROM AUTHOR]

Atherosclerotic plaques are characterized as being vulnerable to rupture based on a series of histologically defined features, including a lipid-rich necrotic core, spotty calcification and ulceration. Existing imaging modalities have limitations in their ability to distinguish between different materials and structural features. We examined whether X-ray spectral photon-counting computer tomography (SPCCT) images were able to distinguish key plaque features in a surgically excised specimen from the carotid artery with comparison to histological images. An excised carotid plaque was imaged in the diagnostic X-ray energy range of 30–120 keV using a small-bore SPCCT scanner equipped with a Medipix3RX photon-counting spectral X-ray detector with a cadmium telluride (CdTe) sensor. Material identification and quantification (MIQ) images of the carotid plaque were generated using proprietary MIQ software at 0.09 mm volumetric pixels (voxels). The plaque was sectioned, stained and photographed at high resolution for comparison. A lipid-rich core with spotty calcification was identified in the MIQ images and confirmed by histology. MIQ showed a core region containing lipid, with a mean concentration of 260 mg lipid/ml corresponding to a mean value of −22HU. MIQ showed calcified regions with mean concentration of 41 mg Ca/ml corresponded to a mean value of 123HU. An ulceration of the carotid wall at the bifurcation was identified to be lipid-lined, with a small calcification identified near the breach of the artery wall. SPCCT derived material identification and quantification images showed hallmarks of vulnerable plaque including a lipid-rich necrotic core, spotty calcifications and ulcerations. [Display omitted] • An excised human carotid plaque was successfully imaged using a preclinical Spectral Photon Counting Computerized Tomography X-ray machine at the clinical X-ray energy range (30–120 KeV). • The spectral photon-counting CT data from a scan of excised human carotid atherosclerotic plaque was used for in silico material identification and quantification that identified lipid-rich and calcium-rich regions within the plaque at 0.1 mm resolution. • Histology confirmed the software-generated material identification spatial assignments within the images corresponded to hallmarks of vulnerable plaque, including lipid-rich necrotic core, spotty calcification and an ulceration associated with lipid and a small calcification. • Use of a filter-hardened X-ray spectrum (30–120 keV) indicates that material identification of lipid-rich regions such as a necrotic lipid core may be achievable with spectral photon-counting CT in a clinical setting. • Comparison of the software-generated material concentration maps to histology of closely associated lipid and calcification indicated that spectral photon-counting CT data will enable lipid measurement even in highly calcified regions of plaque. [ABSTRACT FROM AUTHOR]

Purpose: Photon counting (PC) computed tomography (CT) can provide material-selective CT imaging at low patient dose, but it suffers from suboptimal detector count rate. A dynamic beam attenuator (DBA) can help with the count rate by modulating the x-ray beam intensity such that the low attenuating areas of the patient receive lower exposure, and the detector behind these areas is not overexposed. However, DBA may harden the beam and cause artifacts and errors. This work investigates the positive and negative effects of using DBA in PCCT. Methods: A simple PCCT with a single energy bin, spectroscopic PCCT with 2 and 5 energy bins, and a conventional energy integrating CT were simulated. The simulations were performed with and without using a DBA, at 120 kVp tube voltage and 14 mGy air dose. The DBAs were modeled by soft tissue (ST) equivalent material, iron (Fe), and holmium (Ho) with K-edge located around the middle of the x-ray spectrum. A cylindrical CT phantom and chest phantom with iodine and CaCO 3 contrast elements were used. Image artifacts and quantification errors in general and material decomposed CT images were determined. Results: Simple PCCT exhibited major image artifacts and quantification errors when DBAs were used. The artifacts and errors were decreased with 2-bin spectroscopic PCCT and nearly eliminated with 5-bin spectroscopic CT. The photon starvation noise was higher with Fe-DBA due to strong absorption of lower energy photons. The 5-bin PCCT with ST-DBA and Ho-DBA were nearly free of artifacts and photon starvation noise. The Ho-DBA better preserved the low energy photons due to its K-edge at 55.6 keV, which also decreased beam hardening artifacts and improved material decomposition. The Ho-DBA has a potential to be substantially smaller than Fe-DBA and much smaller than ST-DBA. Conclusion: The DBA fabricated from low Z (such as soft tissue) and K-edge (such as Ho) materials can address the count rate problem of PCCT and provide minimal image artifacts and quantification errors. The DBA fabricated from high Z material with suboptimal K-edge location (such as Fe) can results in major beam hardening artifacts, photon starvation noise, and quantification errors. [ABSTRACT FROM AUTHOR]

In medical dual-energy CT, projection-domain decomposition is typically used to process the acquired two projection datasets. However, when a high atomic-number contrast-agent is introduced, its K-edge discontinuity challenges this method. In this work, we take advantage of the fact that the contrast agent and its solvent are known in advance and propose to first decompose the attenuation coefficient functions of the contrast agent and its solvent to estimate the attenuation contribution of the K-shell photoelectric. Then, the contribution of the K-edge attenuation to the projection data is eliminated and the problem is decomposed to determine the attenuation coefficient functions of the other biological materials. Numerical experiments are shown to verify that the proposed method can give better results with less artifacts than the result of the direct projection-domain decomposition. Moreover, the variances of the decomposition coefficients, which is the index of the noise level of the decomposition coefficient image on the contrast regions, are smaller. [ABSTRACT FROM AUTHOR]

Context: Spectral Computed Tomography (SPCT) acquires energy-resolved projections that can be used to evaluate the concentrations of the different materials of an object. This can be achieved by decomposing the projection images in a material basis, in a first step, followed by a standard tomographic reconstruction step. However, material decomposition is a challenging non-linear ill-posed inverse problem, which is very sensitive to the photonic noise corrupting the data. Methods: In this paper, material decomposition is solved in a variational framework. We investigate two fidelity terms: a weighted least squares (WLS) term adapted to Gaussian noise and the Kullback–Leibler (KL) distance adapted to Poisson Noise. The decomposition problem is then solved iteratively by a regularized Gauss–Newton algorithm. Results: Simulations are performed in a numerical mouse phantom and the decompositions show that KL outperforms WLS for low numbers of photons, i.e., for high photonic noise. Conclusion: In this work, a new method for material decomposition in SPCT is presented. It is based on the KL distance to take into account the noise statistics. This new method could be of particular interest when low-dose acquisitions are performed. [ABSTRACT FROM AUTHOR]

Rationale and Objective This study aimed to investigate the clinical value of spectral computed tomography (CT) in the detection of cholesterol gallstones from surrounding bile. Materials and Methods This study was approved by the institutional review board. The unenhanced spectral CT data of 24 patients who had surgically confirmed cholesterol gallstones were analyzed. Lipid concentrations and CT numbers were measured from fat-based material decomposition image and virtual monochromatic image sets (40–140 keV), respectively. The difference in lipid concentration and CT number between cholesterol gallstones and the surrounding bile were statistically analyzed. Receiver operating characteristic analysis was applied to determine the diagnostic accuracy of using lipid concentration to differentiate cholesterol gallstones from bile. Results Cholesterol gallstones were bright on fat-based material decomposition images yielding a 92% detection rate (22 of 24). The lipid concentrations (552.65 ± 262.36 mg/mL), CT number at 40 keV (−31.57 ± 16.88 HU) and 140 keV (24.30 ± 5.85 HU) for the cholesterol gallstones were significantly different from those of bile (−13.94 ± 105.12 mg/mL, 12.99 ± 9.39 HU and 6.19 ± 4.97 HU, respectively). Using 182.59 mg/mL as the threshold value for lipid concentration, one could obtain sensitivity of 95.5% and specificity of 100% with accuracy of 0.994 for differentiating cholesterol gallstones from bile. Conclusions Virtual monochromatic spectral CT images at 40 keV and 140 keV provide significant CT number differences between cholesterol gallstones and the surrounding bile. Spectral CT provides an excellent detection rate for cholesterol gallstones. [ABSTRACT FROM AUTHOR]

Single-crystal silicon carbide (SiC) is an important semiconductor material used in power electronics. Due to its high hardness and brittleness, SiC is very difficult to machine using mechanical methods. Electrical discharge machining (EDM) has recently garnered extensive research interest as a potential machining method for SiC. However, this technique induces severe subsurface damage on the workpiece. To date, mechanisms leading to EDM-induced subsurface damage in SiC have not been clarified. This study aims to investigate the atomic-scale subsurface damage in SiC induced by EDM using Raman spectroscopy and transmission electron microscopy (TEM). In cross-sectional TEM observations, three regions of subsurface damage were identified, namely, the re-solidified layers, heat-affected zones, and microcracks. It was found that SiC decomposed into silicon and carbon in the re-solidified layers, and the degree of decomposition depended on the discharge energy and workpiece polarity. The re-solidified layer was a mixture of crystalline/amorphous silicon, crystalline/amorphous carbon, and nano-crystalline SiC. The presence of an extremely thin graphite layer was observed in the re-solidified layer. The heat-affected zone remained crystalline but showed a different crystal structure distinct from that of the bulk. [ABSTRACT FROM AUTHOR]

Dual-energy cone-beam computed tomography (DE-CBCT) is a promising imaging technique with foreseeable clinical applications. DE-CBCT images acquired with two different spectra can provide material-specific information. Meanwhile, the anatomical consistency and energy-domain correlation result in significant information redundancy, which could be exploited to improve image quality. In this context, this paper develops the Transformer-Integrated Multi-Encoder Network (TIME-Net) for DE-CBCT to remove the limited-angle artifacts. TIME-Net comprises three encoders (image encoder, prior encoder, and transformer encoder), two decoders (low- and high-energy decoders), and one feature fusion module. Three encoders extract various features for image restoration. The feature fusion module compresses these features into more compact shared features and feeds them to the decoders. Two decoders perform differential learning for DE-CBCT images. By design, TIME-Net could obtain high-quality DE-CBCT images using two complementary quarter-scans, holding great potential to reduce radiation dose and shorten the acquisition time. Qualitative and quantitative analyses based on simulated data and real rat data have demonstrated the promising performance of TIME-Net in artifact removal, subtle structure restoration, and reconstruction accuracy preservation. Two clinical applications, virtual non-contrast (VNC) imaging and iodine quantification, have proved the potential utility of the DE-CBCT images provided by TIME-Net. • First attempt to apply transformer to DE-CBCT. • Exploiting various information via multiple encoders to improve performance. • Obtaining high-quality DE-CBCT images from two complementary quarter-scans is feasible. • Reducing radiation dose and shortening acquisition time for DE-CBCT. [ABSTRACT FROM AUTHOR]

X-ray computed tomography is one of the commonly used diagnostic imaging technique. However, it has some problems which are lack of energy information and high dose exposure. These problems can be solved by using a next generation CT called Photon Counting CT (PC-CT). PC-CT can acquire multi-color images with significantly low radiation dose. The detector system of the PC-CT we propose consists of Multi-Pixel Photon Counter (MPPC) coupled with a high-speed scintillator thus can be made with lower cost and simpler system. In this study, we propose a novel method of material decomposition using the MPPC based PC-CT. Furthermore, we imaged mice with MPPC based PC-CT system as a first attempt. Iodine and gadolinium mixed contrast agents and gold-nanoparticles (AuNPs) were injected when imaging. In addition, the CT images of mice showed bone, lungs, heart, kidneys, and bladder. As a future prospect, we will work on visualizing the pharmacokinetics in mice. [ABSTRACT FROM AUTHOR]

Dual-energy X-ray radiography has become a well-established technique in medical, industrial, and security applications, because of its material or tissue discrimination capability. The main difficulty of this technique is dealing with the materials overlapping problem. When there are two or more materials along the X-ray beam path, its material discrimination performance will be affected. In order to solve this problem, a new dynamic material discrimination algorithm is proposed for dual-energy X-ray digital radiography, which can also be extended to multi-energy X-ray situations. The algorithm has three steps: α -curve-based pre-classification, decomposition of overlapped materials, and the final material recognition. The key of the algorithm is to establish a dual-energy radiograph database of both pure basis materials and pair combinations of them. After the pre-classification results, original dual-energy projections of overlapped materials can be dynamically decomposed into two sets of dual-energy radiographs of each pure material by the algorithm. Thus, more accurate discrimination results can be provided even with the existence of the overlapping problem. Both numerical and experimental results that prove the validity and effectiveness of the algorithm are presented. [ABSTRACT FROM AUTHOR]

Dual energy CT (DECT) scanners are nowadays available in many radiology departments. For radiotherapy purposes, new strategies using DECT imaging are investigated to optimize radiation treatment for multiple steps in the radiotherapy chain. This review describes how DECT based methods can be used for electron density estimation, effective atomic number decomposition and contrast material quantification. Clinical radiotherapy related applications for improved dose calculation accuracy of brachytherapy and proton therapy, metal artifact reduction techniques and normal tissue characterization are also summarized together with future perspectives on the use of DECT for radiotherapy purposes. [ABSTRACT FROM AUTHOR]

Objective: To investigate the value of dual-energy MDCT in spectral imaging in the differential diagnosis of chronic mass-forming chronic pancreatitis (CMFP) and pancreatic ductal adenocarcinoma (PDAC) during the arterial phase (AP) and the pancreatic parenchymal phase (PP).Materials and Methods: Thirty five consecutive patients with CMFP (n=15) or PDAC (n=20) underwent dual-energy MDCT in spectral imaging during AP and PP. Iodine concentrations were derived from iodine-based material-decomposition CT images and normalized to the iodine concentration in the aorta. The difference in iodine concentration between the AP and PP, contrast-to-noise ratio (CNR) and the slope K of the spectrum curve were calculated.Results: Normalized iodine concentrations (NICs) in patients with CMFP differed significantly from those in patients with PDAC during two double phases (mean NIC, 0.26±0.04 mg/mL vs. 0.53±0.02 mg/mL, p=0.0001; 0.07±0.02 mg/mL vs. 0.28±0.04 mg/mL, p=0.0002, respectively). There were significant differences in the value of the slope K of the spectrum curve in two groups during AP and PP (K(CMFP)=3.27±0.70 vs. K(PDAC)=1.35±0.41, P=0.001, and K(CMFP)=3.70±0.17 vs. K(PDAC)=2.16±0.70, p=0.003, respectively). CNRs at low energy levels (40-70 keV) were higher than those at high energy levels (80-40 keV).Conclusion: Individual patient CNR-optimized energy level images and the NIC can be used to improve the sensitivity and the specificity for differentiating CMFP from PDAC by use of dual-energy MDCT in spectral imaging with fast tube voltage switching. [ABSTRACT FROM AUTHOR]

Purpose Spectral Computed Tomography (SCT) systems equipped with photon counting detectors (PCD) are clinically desired, since such systems provide not only additional diagnostic information but also radiation dose reductions by a factor of two or more. The current unavailability of clinical PCDs makes a simulation of such systems necessary. Methods In this paper, we present a Monte Carlo-based simulation of a SCT equipped with a PCD. The aim of this development is to facilitate research on potential clinical applications. Our MC simulator takes into account scattering interactions within the scanned object and has the ability to simulate scans with and without scatter and a wide variety of imaging parameters. To demonstrate the usefulness of such a MC simulator for development of SCT applications, a phantom with contrast targets covering a wide range of clinically significant iodine concentrations is simulated. With those simulations the impact of scatter and exposure on image quality and material decomposition results is investigated. Results Our results illustrate that scatter radiation plays a significant role in visual as well as quantitative results. Scatter radiation can reduce the accuracy of contrast agent concentration by up to 15%. Conclusions We present a reliable and robust software bench for simulation of SCTs equipped with PCDs. [ABSTRACT FROM AUTHOR]

Rationale and Objectives: To evaluate the detectability of urinary calculi on material decomposition (MD) images generated from spectral computed tomography (CT) and identify the influencing factors. Materials and Methods: Forty-six patients were examined with true nonenhanced (TNE) CT and spectral CT urography in the excretory phase. The contrast medium was removed from excretory phase images using water-based (WB) and calcium-based (CaB) MD analysis. The sensitivity for detection on WB and CaB images was evaluated using TNE results as the reference standard. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) on MD images were evaluated. Using logistic regression, the influences of image noise, attenuation, stone size, and patient's body mass index (BMI) were assessed. Threshold values with maximal sensitivity and specificity were calculated by means of receiver operating characteristic analyses. Results: One hundred thirty-six calculi were detected on TNE images; 98 calculi were identified on WB images (sensitivity, 72.06%) and 101 calculi on CaB images (sensitivity, 74.26%). Sensitivities were 76.92% for the 3–5-mm stones and 84.51% for the 5-mm or larger stones on both WB and CaB images but reduced to 46.15% on WB images and 53.85% on CaB images for small calculi (<3 mm). Compared to WB images, CaB images showed lower image noise, higher SNR but similar CNR. Larger stone sizes (both >2.71 mm on WB and CaB) and greater CT attenuation (>280 Hounsfield units [HU] on WB, >215 HU on CaB) of the urinary stones were significantly associated with higher stone visibility rates on WB and CaB images (P ≤ .003). Image noise and BMI showed no impact on the stone detection. Conclusions: MD images generated from spectral CT showed good reliability for the detection of large (>2.71 mm) and hyperattenuating (>280 HU on WB, >215 HU on CaB) urinary calculi. [Copyright &y& Elsevier]

• Computed tomography imaging for cargo/container inspection has been accomplished. • Material decomposition for MeV dual-energy container CT is hard but can be done. • Deep learning is effective on improving the quality of effective atom number image. Traditional inspection technology for cargo or container imaging in customs and harbours is MeV X-ray radiography. The biggest limitation for this technology is the structural overlapping problem, which is inherent to radiography technology. MeV dual energy CT has a major advantage over radiography in that it can provide cross-section image, which is free of the structural overlapping problem. Besides, the recorded dual-energy projection data provides the ability for material decomposition. Electron density image and effective atom number image can be further calculated from the material decomposition coefficient images. However, the quality of effective atom number image can be very poor. The behind reasons are multifaceted. In this paper we proposed an Attenuation Image Referenced (AIR) effective atom number image calculation method for MeV dual-energy container CT imaging by using an image-domain neural network. The network has three channels as input and outputs with the estimated effective atom number image. The input three channels include the low and high-energy attenuation images and the effective atom number image that was directly calculated by using the derived formula. The network utilizes the low and high-energy attenuation image as guidance or reference for the restoration of effective atom number image. The network was trained on synthetic data, which is based on the shape of XCAT model but filled with materials that often appear in security imaging. The trained network also performed well on experimental data, showing the robustness and good generalization ability of the network. The quantitative analysis on the simulation and experimental data that comes from actual MeV dual-energy CT system showed the effectiveness of the proposed Attenuation Image Referenced (AIR) deep learning method for effective atom number image calculation. [ABSTRACT FROM AUTHOR]

Abstract: Energy resolving capabilities of X-ray detectors like the Medipix2 and the upcoming Medipix3 offer access to spectral information which is a new domain of information in medical imaging. In conventional CT of a composite object only the cumulative contribution of all involved materials to the attenuation is measurable, but not how much each material component contributes to this attenuation is measured. Therefore, contrast agent cannot be distinguished from bone or calcifications. The method of material reconstruction exploits the energy information to determine the partial densities of the involved materials using a maximum likelihood approach, i.e. it allows the separation of contrast agent from tissue, bones and calcifications. We have employed the Medipix All Resolution System (MARS) scanner equipped with a Medipix2 MXR and performed a CT scan of a mouse with iodine contrast agent in stomach and bowel. The method allows to separate the iodine contrast agent from all the other absorbing structures. In the iodine image, only the iodine concentration is visible, while the non-iodine (water) image shows all the other tissue structures and bones. The method of material reconstruction was applied to real CT data of a biological sample for the first time. [Copyright &y& Elsevier]

Rationale and Objectives: The identification of body fluids in computed tomography poses a major diagnostic challenge. The chemical composition of body fluids deviates only slightly from water with very similar computed tomographic (CT) values, which typically range from 0 to 100 HU. The aim of this study was to assess physical and chemical properties of different body fluids in an ex vivo setting. Materials and Methods: A total of 44 samples of blood, blood mixed with pus, pus, bile, and urine obtained during diagnostic and therapeutic punctures were scanned at 80 and 140 kV. Data was quantitatively assessed using the spectral ρZ-projection algorithm, which converts dual-energy CT scans into mass density (ρ) and effective atomic number (Zeff. ) information. Results: Attenuation values measured at 80 and 140 kV were largely overlapping. CT values allowed, to some degree, for the differentiation of bile or pus from blood or the blood/pus mixture. By applying the ρZ-projection, most substances, except for urine, were distinguishable with only small standard deviations ranging between 0.003 and 0.007 g/cm3 for mass density and between 0.020 and 0.043 for Zeff. . Conclusion: The ρZ-projection method is suited to quantitatively assess mass density and effective atomic number of ex vivo body fluid samples. In clinical routine, this technique might be useful for identifying unclear fluid collections even in unenhanced computed tomography. [Copyright &y& Elsevier]

Musculoskeletal injuries often induce local accumulation of blood and/or fluid within the bone marrow, which is detected on medical imaging as edema-like marrow signal intensities (EMSI). In addition to its biological effects on post-injury recovery, the displacement of low-attenuating, largely adipocytic marrow by EMSI may introduce errors into quantitative computed tomography (QCT) measurements of bone mineral density (vBMD) and resulting bone stiffness estimates from image-based finite element (FE) analysis. We aimed to investigate the impact of post-injury changes in marrow soft tissue composition on CT-based bone measurements by applying CT imaging at multiple spatial resolutions. To do so, dual energy QCT (DECT) material decomposition was used to detect EMSI in the tibiae of nineteen participants with a recent anterior cruciate ligament tear. We then measured bone density and FE-based apparent modulus within the EMSI region and in a matched volume in the uninjured contralateral knee. Three measurement methods were applied: 1.) standard, QCT density calibration and density-based FEM; 2.) a DECT density calibration that provides density measurements adjusted for marrow soft tissues; and 3.) high-resolution peripheral QCT (HR-pQCT) density and microFE analyses. When measured using standard, single-energy QCT, vBMD and apparent modulus were elevated in the EMSI compared to the contralateral. After adjusting for marrow soft tissue composition using DECT, these measurements were no longer different between the two regions. By allowing for high-resolution, localized density analysis, HR-pQCT indicated that trabecular tissue mineral density was 9 mgHA/cm3 lower, while density of marrow soft tissues was 18 mgHA/cm3 higher, in the EMSI than the contralateral region, suggesting that EMSI have opposite effects on the measured density of trabecular bone and the underlying soft marrow. Thus, after an acute injury, altered composition of marrow soft tissues may artificially inflate overall measurements of bone density and apparent modulus obtained using standard QCT. This can be corrected by accounting for marrow soft tissue attenuation, either by using DECT-based density calibration or HR-pQCT microFE and measurements of local density of trabeculae. [Display omitted] • Post-injury, bone marrow edema increases CT attenuation of marrow soft tissues. • Elevated soft tissue attenuation affects bone density and stiffness measurements. • Dual energy CT measures bone density and stiffness adjusted for marrow composition. • HR-pQCT provides local evaluation of mineralized vs. soft tissue density. [ABSTRACT FROM AUTHOR]

Abstract: In conventional X-ray imaging of a composed object only the cumulative attenuation is accessible, but the contribution of different materials can not be resolved. We consider the object consisting of known basis materials. Our method of material reconstruction provides the areal densities of these basis materials in each pixel of the image. This method uses the different energy dependences of X-ray attenuation of the basis materials to distinguish between them. It obviously requires a spectroscopic detector. Pixelated hybrid semiconductor X-ray detectors such as the Medipix2 suffer from charge sharing between pixels. This leads to a deteriorated energy response of the detector. To overcome this problem, we present two different methods to reconstruct incident X-ray spectra from threshold scans in each pixel. Both methods, namely spectrum stripping and a linear least squares fit, employ MC-simulated energy response functions to monoenergetic irradiation. Both methods work very well. The reconstructed incident spectra allow the application of material reconstruction with Medipix2. We present the material resolved images of a phantom containing aluminum, acrylic glass (PMMA) and iodine. This method allows new applications in non-destructive testing and medical imaging. This work was carried out within the Medipix-collaboration. [Copyright &y& Elsevier]

• Proposed a method to estimate dual-energy CT images from traditional single energy CT image together with a single view projection at a different energy level. • Deep-learning algorithm is used to perform material-decomposition-like operations. • The approach can provide superior material-specific images with significantly reduced noise compared with standard DECT imaging. [Display omitted] Dual-energy computed tomography (DECT) is of great significance for clinical practice due to its huge potential to provide material-specific information. However, DECT scanners are usually more expensive than standard single-energy CT (SECT) scanners and thus are less accessible to undeveloped regions. In this paper, we show that the energy-domain correlation and anatomical consistency between standard DECT images can be harnessed by a deep learning model to provide high-performance DECT imaging from fully-sampled low-energy data together with single-view high-energy data. We demonstrate the feasibility of the approach with two independent cohorts (the first cohort including contrast-enhanced DECT scans of 5753 image slices from 22 patients and the second cohort including spectral CT scans without contrast injection of 2463 image slices from other 22 patients) and show its superior performance on DECT applications. The deep-learning-based approach could be useful to further significantly reduce the radiation dose of current premium DECT scanners and has the potential to simplify the hardware of DECT imaging systems and to enable DECT imaging using standard SECT scanners. [ABSTRACT FROM AUTHOR]

Neutron cross-sections have an energy dependence which is uniquely varying according to isotopic composition. By measuring the local neutron beam attenuation of an object as a function of energy, this can be related to its elemental composition. Therefore material decomposition measurements using a deuterium-deuterium (D-D) compact fast neutron generator in an energy-selective fashion have been carried out. Previous work described determining setup-specific effective reference cross-section data for homogeneous samples. This work presents first tests of this non-destructive analysis principle in a 2D computed tomography (CT) setup using heterogeneous samples, based on the aforementioned reference cross-sections. Measurement data were collected at five angles around the neutron generator corresponding to five effective energies. At each angle a CT reconstruction was performed to obtain energy-specific CT images. These CT images were use to calculate an elemental decomposition of the samples of interest. Good agreement between theory and experiment was found for most of the samples, while the more challenging cases highlighted limitations of the technique. These first tomographic measurements are discussed along with an outlook for further developments of the technique. • Material decomposition around a deuterium-deuterium neutron generator. • Two heterogeneous test phantoms were imaged with a computed tomog- raphy setup. • Based on reference cross-section data, their element content was resolved. • Compared to expected composition, the agreement was found to be good. [ABSTRACT FROM AUTHOR]

This work aims to describe experimental efforts to further development of an energy-selective based interrogation technique using a deuterium-deuterium (D-D) fast neutron generator. In this approach, elemental decomposition of a sample can be performed. Limitations of the technique include uncertainties related to the neutron spectrum and the actual elemental cross-section data. To circumvent these challenges, a calibration method using well-characterized homogeneous reference samples has been implemented. For this procedure, twenty-two reference samples were used to produce setup-specific reference data which can afterwards be used for elemental decomposition measurements using fast neutrons. Data for individual elements were obtained by measuring pure elemental samples or by subtracting elemental sample data from compound samples. Ultimately full cross-section characterization was obtained for nineteen individual elements. Although limitations to this approach were found, the general agreement between simulation and experiment was good. This technique, the measured results, and how it can be used for fast neutron elemental decomposition are described. • Material decomposition around a deuterium-deuterium neutron generator. • Spectrum and cross-section uncertainties led to a calibration procedure. • Nineteen individual elements were isolated, and their cross-section characterized. • Even considering experimental uncertainty, the agreement was found to be good. [ABSTRACT FROM AUTHOR]

Dual-energy X-ray inspection can provide material-decomposed transmission information of the imaged object enhancing observer's inspection performance. The material decomposition algorithms have an inherent noise amplification problem due to the multiplication and division operations during dual-energy image processing, which often degrades the material decomposition performance. In this work, we propose an effective and reliable clustering algorithm to resolve the noise amplification problem. The proposed method utilizes a k-means cluster algorithm as a base method with Gaussian noise of the data considered. In addition, we added a probabilistic regularization that can effectively relieve the decision challenge of the initial number of clusters. We have applied the proposed denoising method to the prototype X-ray dual-energy inspection system with an empirical calibration method that we have previously proposed. The proposed algorithm showed substantial reduction of noise in the test container projections and accordingly improved the material decomposition accuracy. Moreover, we performed quantitative analysis of the empirical calibration method with and without the cluster-based noise reduction algorithm and also on the conventional calibration method for material decomposition. It showed that the empirical calibration method with the noise reduction algorithm is robust against the thickness variation of the target object. The results suggest that the proposed noise reduction algorithm with the empirical calibration method is a promising solution to dual-energy X-ray material decomposition for inspection. • We developed a noise reduction algorithm for dual-energy X-ray imaging. • A spatially constrained k -means clustering algorithm with a probabilistic regularization has been proposed. • The performance of the proposed noise reduction method was superior to the conventional noise reduction algorithm in the experimental results. [ABSTRACT FROM AUTHOR]