Skip to main content
NCBI home page
Medical Physics logoLink to Medical Physics
. 2019 Jul 5;46(9):4105–4115. doi: 10.1002/mp.13668

Feasibility of multi‐contrast imaging on dual‐source photon counting detector (PCD) CT: An initial phantom study

Shengzhen Tao 1, Kishore Rajendran 1, Cynthia H McCollough 1, Shuai Leng 1,✉
PMCID: PMC6857531  NIHMSID: NIHMS1054388  PMID: 31215659

Abstract

Purpose

Photon‐counting‐detector‐computed tomography (PCD‐CT) allows separation of multiple, simultaneously imaged contrast agents, such as iodine (I), gadolinium (Gd), and bismuth (Bi). However, PCDs suffer from several technical limitations such as charge sharing, K‐edge escape, and pulse pile‐up, which compromise spectral separation of multi‐energy data and degrade multi‐contrast imaging performance. The purpose of this work was to determine the performance of a dual‐source (DS) PCD‐CT relative to a single‐source (SS) PCD‐CT for the separation of simultaneously imaged I, Gd, and Bi contrast agents.

Methods

Phantom experiments were performed using a research whole‐body PCD‐CT and head/abdomen‐sized phantoms containing vials of different I, Gd, Bi concentrations. To emulate a DS‐PCD‐CT, the phantoms were scanned twice on the SS‐PCD‐CT using different tube potentials for each scan. A tube potential of 80 kV (energy thresholds = 25/50 keV) was used for low‐energy tube, while the high‐energy tube used Sn140 kV (Sn indicates tin filter) and thresholds of 25/90 keV. The same phantoms were scanned also on the SS‐PCD‐CT using the chess acquisition mode. In chess mode, the 4 × 4 subpixels within a macro detector pixel are split into two sets based on a chess‐board pattern. With each subpixel set having two energy thresholds, chess mode allows four energy‐bin data sets, which permits simultaneous multi‐contrast imaging. Because of this design, only 50% area of each detector pixel is configured to receive photons of a pre‐defined threshold, leading to 50% dose utilization efficiency. To compensate for this dose inefficiency, the radiation dose for this scan was doubled compared to DS‐PCD‐CT. A 140 kV tube potential and thresholds = 25/50/75/90 keV were used. These settings were determined based on the K‐edges of Gd, and Bi, and were found to yield good differentiation of I/Gd/Bi based on phantom experiments and other literature. The energy‐bin images obtained from each scan (scan pair) were used to generate I‐, Gd‐, Bi‐specific image via material decomposition. Root‐mean‐square‐error (RMSE) between the known and measured concentrations was calculated for each scenario. A 20‐cm water cylinder phantom was scanned on both systems, which was used for evaluating the magnitude of noise, and noise power spectra (NPS) of I/Gd/Bi‐specific images.

Results

Phantom results showed that DS‐PCD‐CT reduced noise in material‐specific images for both head and body phantoms compared to SS‐PCD‐CT. The noise level of SS‐PCD was reduced from 2.55 to 0.90 mg/mL (I), 1.97 to 0.78 mg/mL (Gd), and 0.85 to 0.74 mg/mL (Bi) using DS‐PCD. NPS analysis showed that the noise texture of images acquired on both systems is similar. For the body phantom, the RMSE for SS‐PCD‐CT was reduced relative to DS‐PCD‐CT from 10.52 to 2.76 mg/mL (I), 7.90 to 2.01 mg/mL (Gd), and 1.91 to 1.16 mg/mL (Bi). A similar trend was observed for the head phantom: RMSE reduced from 2.59 (SS‐PCD) to 0.72 (DS‐PCD) mg/mL (I), 2.02 to 0.58 mg/mL (Gd), and 0.85 to 0.57 mg/mL (Bi).

Conclusion

We demonstrate the feasibility of performing simultaneous imaging of I, Gd, and Bi materials on DS‐PCD‐CT. Under the condition without cross scattering, DS‐PCD reduced the RMSE for quantification of material concentration in relative to a SS‐PCD‐CT system using chess mode.

Keywords: dual‐source CT, multi‐contrast imaging, multi‐energy CT, photon counting detector, spectral CT

1. Introduction

The scintillating x‐ray detectors used in clinical computed tomography (CT) systems integrate the absorbed energy from all detected photons. Multi‐energy acquisitions, such as required for dual‐energy (DE) CT, are accomplished by a number of approaches, including those that use two tube potentials, that is, dual‐source (DS) CT or fast kV switching, and those that use a single tube potential, that is, split‐beam‐filter or dual‐layer‐detector methods.1 Photon counting detector (PCD) CT is an emerging technology that has recently been used to image human subjects.2, 3, 4, 5 Compared to conventional energy‐integrating‐detector (EID) technology, PCDs have demonstrated several advantages from their direct conversion of x‐ray photons to electronic signals, such as reduced electronic noise due to energy thresholding,6 improved iodine contrast by removing the energy weighting in the detected signals,7, 8 and dose‐efficient high‐resolution imaging.9 Another unique feature of PCD‐CT is its capability to discriminate the energy of incident photons, which makes it intrinsically suitable for multi‐energy CT applications.10 With PCD‐CT, a series of energy‐bin images can be generated from a single data acquisition, with each energy‐bin image representing the photons detected within a user‐defined energy window. This allows PCD‐CT to measure the change in x‐ray attenuation due to the K‐edges present in materials such as gadolinium (Gd) and bismuth (Bi).11, 12, 13, 14, 15, 16, 17, 18, 19, 20

This energy‐discriminating capability of PCD‐CT facilitates the separation of simultaneously imaged contrast materials. This is especially relevant for imaging of novel nanoparticle‐based contrast agents having a detectable K‐edge in the diagnostic x‐ray energy range.21, 22, 23, 24 After tagging them with relevant molecules, these nanoparticles can be used as probes to selectively image a specific target or pathology of interest, such as a vulnerable plaque. Multiple nanoparticles (e.g., iodine (I), Gd and Bi) can be administered to simultaneously image different targets or molecular pathways of interest. The use of these targeted contrast agents may extend the frontier of CT beyond conventional anatomical imaging toward molecular imaging to eventually enable novel CT applications. In addition, the use of multiple contrast agents may be relevant for clinical applications that require multi‐phase acquisitions,11, 14, 19, 20, 25, 26, 27 where repeated scans are required to capture the contrast enhancement at different time points, which can lead to increased radiation dose.28 With multi‐contrast imaging, a single scan is acquired after injection of different contrast agents at different time points. Then, a series of contrast‐specific images can be generated retrospectively, using material decomposition, to depict each contrast enhancement phase.11, 22, 29, 30 The use of multiple contrast agents can therefore reduce the number of scans and might present an opportunity to reduce patient radiation dose.

The performance of multi‐contrast imaging relies on the degree to which the acquired multi‐energy datasets represent unique distributions of photon energies. Limitations of PCD‐CT technology, such as K‐escape, charge sharing, and pulse pileup, cause the energy response of each energy bin to deviate considerably from the ideal response.31, 32, 33, 34 The resulting imperfect separation of the acquired energy distributions (i.e., the presence of spectral overlap) degrades the performance of multi‐contrast imaging. With EID‐based CT, DE datasets can be acquired by using two sets of tube/detector pairs (i.e., DSCT), with each pair operating at different tube potentials.1 An additional beam filter, such as tin, can be added to one of x‐ray tube to improve spectral separation.35 The same approach can be utilized in PCD‐CT, yielding a DS‐PCD‐CT system architecture. Previous works based on numerical simulation and phantom experiments have shown that such a DS‐PCD‐CT may provide benefit for conventional DE imaging applications such as iodine‐ and water‐based material decomposition.36, 37 Compared to two‐material decomposition, multi‐contrast imaging is a more challenging task with images more susceptible to imperfect PCD spectral response, as well as noise amplification during material decomposition. By combining DS architecture and PCD technology, the spectral separation among the acquired energy‐bin datasets may be improved, which can potentially benefit multi‐contrast imaging. The purpose of this work was to demonstrate the feasibility of performing multi‐contrast imaging of I, Gd, and Bi contrast agents on a DS‐PCD‐CT platform based on phantom experiments, and compare its performance to an existing single‐source PCD‐CT using chess acquisition mode. A DS‐PCD‐CT was emulated in this study by scanning the same object twice with different tube potentials due to the absence of a true DS‐PCD system. Because of this setup, the effect of cross scattering between the two subsystems will not be included. The cross scattering can affect the spectral separation of the data acquired on a DS‐PCD system. However, the dual‐scan setup still allows us to demonstrate the key concept and to answer the question regarding the performance of DS‐PCD using realistic PCD units under the condition without cross scattering. This is the first step toward the development and investigation of such systems and has important implication and meaning for showing the benefits of DS‐PCD system.

2. Materials and methods

2.1. Research Whole‐Body PCD‐CT

Phantom experiments involving three contrast materials including I, Gd, and Bi were performed on a research whole‐body PCD‐CT (Somatom CounT, Siemens Healthcare GmbH, Forchheim, Germany).8 This system is a second generation DSCT (Somatom Definition Flash, Siemens Healthcare GmbH), where the detector array on one of the two tube/detector pairs was replaced with a PCD array. The x‐ray source associated with the PCD array is the same as that associated with the EID array. Additionally, a 0.4 mm tin (Sn) filter can be inserted into the x‐ray beam of the tube associated with the PCD array. A true DS‐PCD‐CT system architecture is emulated on this single‐source (SS) PCD‐CT by scanning the same phantom twice using two different tube potential/beam filter combinations.

Currently, four data acquisition modes have been implemented on the SS‐PCD‐CT system that is, macro, chess, sharp, and ultra‐high resolution modes.9 Different acquisition modes have different numbers of energy thresholds, minimum detector readout pixel sizes, and longitudinal collimations. For the purpose of this work, we focus on the macro and chess mode, which have the same detector pixel size at isocenter (0.5 mm) and the same longitudinal coverage of 32 × 0.5 = 16 mm. The macro mode allows two energy thresholds, which can be used to generate two unique energy‐bin datasets. The chess mode splits a detector pixel into two sets of interleaved subpixels based on a chess‐board pattern. With each of the two subpixels sets having two energy thresholds, the chess mode allows a total of four unique energy‐bin data sets.8 It is therefore capable of multi‐contrast imaging using I, Gd, and Bi. However, due to the chess‐board pattern design, only half of the detector pixel area in chess mode is configured to receive photons of a pre‐defined threshold, leading to 50% dose utilization efficiency. As a result, the radiation dose applied in the chess mode should be doubled to get the same dose being used for image formation as a true PCD with four energy thresholds. Note that the chess mode would not be used clinically and was created on this research system simply to allow acquisition of four energy bins in a single scan with two counters for each subpixel. A detector with four energy thresholds in every detector pixel would not suffer from this dose penalty; however, the spectral overlap between each energy bin would be similar. Charge sharing effect for the chess mode used here is equivalent to a true 4‐threshold detector despite their difference in readout electronic setup. This is because charge sharing happens before the readout electronics. Each detector subpixel will receive the charge shared from neighboring subpixels, with the amount of charge solely determined by physical interaction and spatial cross talk.

2.2. Experiments

Numerical simulation was first performed to compare the spectral separation between the SS‐PCD‐CT chess mode and an emulated DS‐PCD‐CT macro mode using vendor‐provided simulation tool developed for the research whole‐body PCD‐CT system.38 The simulation tool can model the deposition of x‐ray energy in the direct converting material (e.g., CdTe) due to various physical interactions including Rayleigh scatter, Compton scatter, and photoelectric effect. It can also simulate the propagation of the charge clouds and consider imperfect detector responses due to charge sharing, fluorescence x‐rays, and cross‐talk between different detector pixels, as well as simulate the effect of energy thresholding and electronic noise.39 As such, the energy spectra of transmitted x‐ray beam were generated using the simulation tool after accounting for both input x‐ray spectrum and detector response. SS‐PCD‐CT chess mode acquisition was performed with 140 kV tube potential to provide the widest range of energy thresholds, which were set as 25, 50, 75, 90 keV, yielding a total of four energy‐bin datasets. The 25 keV threshold was used to reject electronic noise, while the 50 and 90 keV thresholds were used to capture the K‐edge of Gd and Bi, respectively. Setting thresholds based on K‐edges helps improving the differentiation of these contrast agents, as shown in various studies.13, 14, 22 The 75 keV threshold was selected to balance the number of photons within adjacent energy bins. Based on previous phantom experiments and other literature,11 these thresholds yield good performance for differentiating I, Gd, and Bi contrast agents. The 140 kV tube potential allows more photons to be included in the highest energy bin compared to a lower tube potential such as 120 kV.

For the emulated DS‐PCD‐CT, tube potentials of 80 and 140 kV were used to maximize the energy separation between the low‐ and high‐energy scans. A 0.4 mm Sn filter was used for the 140 kV scan to further improve energy separation. Macro mode, with its two energy thresholds, was used for the emulated DS‐PCD‐CT. The energy thresholds were set at 25 and 50 keV for the 80 kV scan and 25 and 90 keV for 140 kV + Sn scan, yielding a total of four energy‐bin images, the same as for the SS‐PCD‐CT chess mode acquisition. The detected energy spectra after passing through 20 and 30 cm diameter water phantoms were generated using vendor‐provided software,38 which simulated the attenuation of the head and abdomen phantoms, respectively.

Phantom experiments were then performed to evaluate the performance of multi‐contrast imaging using DS‐PCD‐CT. To emulate head and abdominal CT scans, a 20‐cm diameter circular head phantom made from water‐equivalent material (Gammex Inc., Middletown, WI, USA), as well as an abdominal‐shaped (30 × 25 cm) water phantom, was scanned on the PCD‐CT system using the tube potentials and acquisition modes described above. Vials containing I, Gd and Bi solutions of different concentrations were placed within each phantom. Iodine solutions with concentrations of 5 and 10 mg/mL were generated by diluting a clinical iodine contrast agent (Omnipaque 300, GE Healthcare, Milwaukee, WI, USA) with water. Gadolinium solutions of 4 and 8 mg/mL concentration were prepared from a clinical Gd contrast agent (Dotarem, Guerbet, Princeton, NJ, USA). Finally, 5 and 10 mg/mL Bi solutions were prepared by diluting Pepto Bismol solution (Pepto Bismol, Procter and Gamble, Cincinnati, OH, USA). In addition, two multi‐contrast mixture solutions were also prepared, with one containing 5 mg/mL of I and 4 mg/mL of Gd, and the other containing 5 mg/mL of Bi, 4 mg/mL of Gd and 5 mg/mL of I.

To emulate a DS‐PCD‐CT scan, the head phantom was scanned twice on the SS‐PCD‐CT system in macro mode using 0.55 spiral pitch, collimation = 32×0.5 mm, rotation time = 1 s, full scan, 474 mAs at 80 kV and 360 mAs at Sn140 kV, and CTDIvol, 16cm of 26.35 mGy at 80 kV and 26.45 mGy at Sn140 kV. The energy thresholds were set at 25/50 keV for the 80 kV scan and 25/90 keV for the Sn140 kV scan. The total radiation dose for the emulated DS‐PCD‐CT scan (52.8 mGy CTDIvol, 16cm) was similar to that of a clinical head CT scan. The head phantom was also scanned on the SS‐PCD‐CT with the chess mode using 0.55 spiral pitch, collimation = 32×0.5 mm, rotation time = 1 s, full scan, 416 mAs, and a CTDIvol, 16cm of 106 mGy. The tube potential and energy thresholds were the same as used in simulation. Note that, because of the 50% dose efficiency of the chess mode, the SS‐PCD‐CT scan was performed at twice the radiation dose compared to the DS‐PCD‐CT to compensate for this dose inefficiency.

Next, the abdominal water tank phantom containing I, Gd, and Bi solutions was scanned using the emulated DS‐PCD‐CT protocol (80/Sn140 kV, collimation = 32×0.5 mm, rotation time = 0.5 s, full scan, 316/200 mAs, CTDIvol, 32cm = 7.0/6.95 mGy, energy thresholds = 25/50 keV and 25/90 keV for the 80 and Sn140 scans, respectively). The total radiation dose level (13.95 mGy CTDIvol, 16cm) of the DS‐PCD scan was similar to that of a routine clinical abdominal CT scan. The same body phantom was scanned using the SS‐PCD chess mode abdominal protocol with 140 kV, energy thresholds = 25/50/75/90 keV, collimation = 32×0.5 mm, rotation time = 0.5 s, full scan, 252 mAs, CTDIvol, 32cm = 27.91 mGy. Again, the total radiation dose of the SS‐PCD‐CT chess mode acquisition was twice as that of DS‐PCD‐CT.

A 20 cm diameter cylindrical water phantom was further scanned on DS‐PCD‐CT and SS‐PCD‐CT using the same acquisition settings as the head phantom experiment to evaluate the noise power spectra. Each acquisition was repeated three times.

2.3. Data processing

For each acquisition, the energy‐bin images were reconstructed with a weighted filtered back projection algorithm40 using a medium‐smooth quantitative (D30) kernel available with a vendor‐provided off‐line reconstruction package (Recon CT, Siemens Healthcare GmbH). Images were reconstructed with a field‐of‐view of 25 cm, and a 512 × 512 matrix size. The image thickness for the head and body scans were 2 and 5 mm, respectively. Next, material decomposition was performed using an image domain least‐square–based approach with volume conservation constraint, assuming I, Gd, Bi, and water as basis materials. The mean values and standard deviations of material concentrations were measured for each vial using a 10‐mm diameter circular region‐of‐interest. The root‐mean‐square‐error (RMSE) for quantification of material concentration between the known and measured concentrations, which includes the effects of bias and noise, was also calculated for comparison between the SS‐PCD chess mode scans and the emulated DS‐PCD macro‐mode scans. The RMSE calculation included all pixels within the ROIs in different contrast vials.

The noise power spectra (NPS) of each material‐specific images decomposed using SS‐PCD and DS‐PCD images were generated from the water phantom scans. To calculate NPS, a total of 40 square ROIs (9 × 9 mm2) located on a 70 mm radius circle were extracted from the I‐, Gd‐, and Bi‐specific images obtained from the water scan. The NPS curves of each material‐specific image were then generated following the method described by Siewerdsen et al.41, and then averaged among three repeated scans. The image noise level of each material images was also calculated as the standard deviation measured using a cylindrical ROI (diameter = 40 mm, length = 10 mm) in the center of the phantom.

3. Results

3.1. Simulations

Figures 1(a) and 1(c) shows the energy spectra for the four energy bins of a SS‐PCD chess mode acquisition after 20 and 30 cm of water attenuation, respectively. Note that the long high‐energy tails of the lower energy‐bin spectra (e.g., 25–50 and 50–75 keV bins) are due to the imperfect detector response33 and lead to substantial overlap between different energy bins. The mean energies of the four bins are 71.8, 75.5, 92.5, 109.3 keV with 20 cm of water attenuation, and are 76.0, 78.1, 93.6, 110.1 keV with 30 cm of water attenuation. In both cases, the mean energies of the 25–50, 50–75, and 75–90 keV energy bins exceed the supposed upper energy limit of the bins, with the mean energy of 25–50 keV bin being almost equivalent to that of the 50–75 keV bin.

Figure 1.

Figure 1

Open in a new tab

Energy spectra of four energy bins for: (a, c) single‐source (SS)‐PCD chess mode scans with energy thresholds of 25, 50, 75, and 90 keV after passing through 20 cm (a) and 30 cm (c) of water attenuation, and (b, d) emulated dual‐source (DS)‐PCD macro‐mode scans using 80 kV (energy thresholds = 25 and 50 keV) and 140 kV with tin filter (energy thresholds = 25 and 90 keV) for 20 cm (b) and 30 cm (d) of water attenuation. The mean energies of each bin (Em) are provided in the figure legend. [Color figure can be viewed at wileyonlinelibrary.com]

Figures 1(b) and 1(d) shows the energy spectra of the four energy bins for an emulated DS‐PCD macro‐mode acquisition after 20 and 30 cm of water attenuation, respectively. The mean energy of the four energy bins are 57.2, 63.9, 87.7, 110.9 keV and 59.2, 64.7, 90.3, 111.7 keV for 20 and 30 cm of water attenuation, respectively. The emulated DS‐PCD approach substantially reduced the mean energies of the 25–50, 50–75, and 75–90 keV energy bins. For the 25–50 keV energy bin, for example, the mean energy is 71.8 keV for SS‐PCD chess mode approach vs 57.2 keV using the DS‐PCD macro‐mode approach. The differences in mean energies between adjacent energy bins are summarized in Table 1, which shows that the energy separation between adjacent energy bins increased using the DS‐PCD approach.

Table 1.

The differences in mean energies between adjacent energy bins using single‐source PCD chess mode and dual‐source PCD macro‐mode acquisitions after 20 and 30 cm of water attenuation.

  Difference in mean energies of Bins 1 and 2 (keV) Difference in mean energies of Bins 2 and 3 (keV) Difference in mean energies of Bins 3 and 4 (keV)
20 cm (Head)
SS‐PCD 3.7 17.0 16.8
DS‐PCD 6.7 23.8 23.2
30 cm (Body)
SS‐PCD 2.1 15.5 16.5
DS‐PCD 5.5 25.6 21.4
Open in a new tab

3.2. Experiments

Figure 2 shows the material‐specific images for I, Gd, and Bi in the head phantom for the SS‐PCD chess mode and emulated DS‐PCD macro‐mode scan data. The low‐energy threshold (TL) image of the head phantom is also shown. For the SS‐PCD‐CT, noise in the Bi image is lower than Gd/I images. This is due to better energy separation in higher energy photon range. As shown in Table 1, the difference of mean energy between bins 1 and 2 is 3.7 keV while that between bins 3 and 4 is 16.8 keV, using SS‐PCD‐CT. The image noise in Gd‐ and I‐specific images generated from DS‐PCD‐CT is reduced compared to that generated from SS‐PCD‐CT. This is in part due to the improved energy separation for the lower energy bins using the DS‐PCD approach (Table 1) compared to SS‐PCD‐CT. As shown in Fig. 3, the I, Gd, and Bi mass concentration RMSE values for the DS‐PCD approach (0.72, 0.58, and 0.57 mg/mL) were reduced compared to the 2 × dose SS‐PCD approach (1.93, 1.50, and 0.64 mg/mL), yielding a 63%, 61%, 10% reduction for I, Gd, and Bi mass quantification, respectively. Table 2 summarizes the mean values and standard deviations of material concentrations measured for each vial. Both techniques showed low quantification bias comparing between the mean concentration values and the corresponding true values, while the DS‐PCD results showed consistently lower noise in different material‐specific images compared to SS‐PCD.

Figure 2.

Figure 2

Open in a new tab

The iodine (I, a and d), gadolinium (Gd, b and e), and Bismuth (Bi, c and f) images generated from the head phantom scans using a single‐source PCD chess mode acquisition (a–c) and dual‐source PCD macro‐mode acquisition (d–f). The CT image (energy threshold‐low image) of the phantom setup is also shown. There are a total of eight vials within the phantom: two containing I (5, 10 mg/mL), two containing Gd (4, 8 mg/mL), two containing Bi (5, 10 mg/mL), as well as two containing mixtures (M1: 5 mg/mL I + 4 mg/mL Gd; M2: 5 mg/mL I + 4 mg/mL Gd + 5 mg/mL Bi). Window/level = 600/100 HU (CT) and 15/7.5 mg/mL (I, Gd, Bi images). [Color figure can be viewed at wileyonlinelibrary.com]

Figure 3.

Figure 3

Open in a new tab

The root‐mean‐square‐error (RMSE) of mass concentration values for iodine (I), gadolinium (Gd), and bismuth (Bi) in the head phantom images for data acquired using a single‐source (SS)‐PCD chess mode scan and dual‐source (DS)‐PCD macro‐mode scans. The RMSE calculation included all pixels within the ROIs in different contrast vials. [Color figure can be viewed at wileyonlinelibrary.com]

Table 2.

Iodine (I), gadolinium (Gd), and bismuth (Bi) mass concentration measurements (mean ± standard deviation) for each vial in the head phantom for data acquired using a single‐source (SS)‐PCD chess mode scan and dual‐source (DS)‐PCD macro‐mode scans.

  I 5 mg/mL I 10 mg/mL Gd 4 mg/mL Gd 8 mg/mL Bi 5 mg/mL Bi 10 mg/mL Mixture 1 (I 5 mg/mL + Gd 4 mg/mL) Mixture 2 (I 5 mg/mL + Gd 4 mg/mL + Bi 5 mg/mL)
SS‐PCD with double radiation dose (CTDIvol,16cm = 106 mGy) 5.2 ± 1.9 10.0 ± 1.7 3.5 ± 1.4 8.0 ± 1.4 4.9 ± 0.6 10.1 ± 0.7 5.4 ± 2.0 (I) 3.9 ± 1.5 (Gd)

4.9 ± 2.0 (I)

4.1 ± 1.6 (Gd)

5.2 ± 0.6 (Bi)
DS‐PCD (CTDIvol,16cm = 53 mGy) 5.1 ± 0.7 9.9 ± 0.8 4.0 ± 0.6 8.0 ± 0.6 4.8 ± 0.5 10.1 ± 0.6 5.2 ± 0.7 (I) 4.0 ± 0.6 (Gd)

5.0 ± 0.7 (I)

3.9 ± 0.6 (Gd)

5.2 ± 0.5 (Bi)
Open in a new tab

The I, Bi, and Gd material images generated from the body phantom scans performed using SS‐PCD and DS‐PCD are shown in Fig. 4, along with the corresponding CT (TL) image. Compared to the head mode scan in Fig. 2, the image noise in the I‐ and Gd‐specific images of the body phantom are substantially higher, which obscured the delineation of the phantom inserts. This is in part because of the narrower difference in energy spectra of the lower energy bins (e.g., 2.1 keV between bin 1 and bin2) when imaging a larger object. Note that a larger object could further harden the x‐ray beam, reduce the separation between lower energy bins, and degrade the material decomposition performance. On the other hand, the multi‐contrast imaging performance is substantially improved using the DS‐PCD approach. The material quantification RMSE values of images acquired from DS‐PCD scan and SS‐PCD scan with double radiation dose are summarized in Fig. 5. Compared to the head phantom scan, the Gd and I quantification RMSEs are increased by four times, and are 9.83, 6.84, 1.55 mg/mL for I, Gd and Bi, respectively. Using the DS‐PCD approach, the RMSE values of I, Gd, and Bi quantification are reduced to 2.76, 2.01, and 1.16 mg/mL — a 71%, 71%, and 25% reduction, respectively. The I, Gd, and Bi concentrations measured from each phantom inserts are summarized in Table 3, which shows that both measurement bias and noise are reduced using DS‐PCD compared to the SS‐PCD chess mode, especially for the Gd and I images. More material cross‐talk between Gd and I signals can be observed in the SS‐PCD results compared to DS‐PCD (see Mixtures 2 in Table 3), as shown by the more positively biased I signal (+2.2 mg/mL for SS‐PCD; +0.5 mg/mL for DS‐PCD) and negatively biased Gd signal (−1.4 mg/mL for SS‐PCD; −0.5 mg/mL for DS‐PCD).

Figure 4.

Figure 4

Open in a new tab

The iodine (I, a and d), gadolinium (Gd, b and e), and Bismuth (Bi, c and f) images generated from the body phantom scans using a single‐source (SS)‐PCD chess mode acquisition (a–c) and dual‐source(DS)‐PCD macro‐mode acquisition (d–f). The CT image (energy threshold‐low image) of the phantom setup is also shown. There are a total of eight vials within the phantom: two containing I (5, 10 mg/mL), two containing Gd (4, 8 mg/mL), two containing Bi (5, 10 mg/mL), as well as two containing mixtures (M1: 5 mg/mL I + 4 mg/mL Gd; M2: 5 mg/mL I + 4 mg/mL Gd + 5 mg/mL Bi). Window/level = 600/100 HU (CT) and 15/7.5 mg/mL (I, Gd, Bi images). [Color figure can be viewed at wileyonlinelibrary.com]

Figure 5.

Figure 5

Open in a new tab

The root‐mean‐square‐error (RMSE) of mass concentration values for iodine (I), gadolinium (Gd), and bismuth (Bi) in the body phantom images for data acquired using a single‐source (SS)‐PCD chess mode scan and dual‐source (DS)‐PCD macro‐mode scans. The RMSE calculation included all pixels within the ROIs in different contrast vials. [Color figure can be viewed at wileyonlinelibrary.com]

Table 3.

Iodine (I), gadolinium (Gd), and bismuth (Bi) mass concentration measurements (mean ± standard deviation) for each vial in the body phantom for data acquired using a single‐source (SS)‐PCD chess mode scan and dual‐source (DS)‐PCD macro‐mode scans.

  I 5 mg/mL I 10 mg/mL Gd 4 mg/mL Gd 8 mg/mL Bi 5 mg/mL Bi 10 mg/mL Mixture 1 (I 5 mg/mL + Gd 4 mg/mL) Mixture 2 (I 5 mg/mL + Gd 4 mg/mL + Bi 5 mg/mL)
SS‐PCD with double radiation dose (CTDIvol,32cm = 28 mGy) 5.9 ± 9.1 10.3 ± 10.7 3.1 ± 7.2 8.8 ± 6.2 4.7 ± 1.3 10.4 ± 1.7 5.5 ± 10.0 (I) 3.6 ± 7.2 (Gd)

7.2 ± 9.2 (I)

2.6 ± 6.5 (Gd)

5.3 ± 1.4 (Bi)
DS‐PCD (CTDIvol,32cm = 14 mGy) 5.5 ± 3.0 10.0 ± 2.7 3.9 ± 1.8 8.2 ± 1.9 5.1 ± 1.1 10.1 ± 1.1 5.4 ± 2.7 (I) 3.6 ± 2.2 (Gd)

5.5 ± 2.4 (I)

3.5 ± 2.0 (Gd)

5.5 ± 1.2 (Bi)
Open in a new tab

Figure 6 shows the NPS curves of I‐, Gd‐, and Bi‐specific images generated from water scans performed on SS‐PCD‐CT and DS‐PCD‐CT. The NPS of the two systems have similar shapes between each other, while the amplitude of NPS is reduced using DS‐PCD. This indicates that the noise texture is similar between the two systems. The noise levels in the I‐, Gd‐, Bi‐specific images were reduced from 2.55, 1.97, 0.85 mg/mL using SS‐PCD to 0.90, 0.78, 0.74 mg/mL using DS‐PCD.

Figure 6.

Figure 6

Open in a new tab

The noise power spectra of iodine (I)‐, gadolinium (Gd)‐, and bismuth (Bi)‐ specific images generated from data acquired on SS‐PCD‐CT and DS‐PCD‐CT. [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

In this work, we evaluated the ability of a DS‐PCD‐CT acquisition strategy to improve the energy separation in multi‐energy CT datasets compared to a SS‐PCD‐CT approach that acquired four energy‐bin datasets in a single scan. The DS‐PCD approach was emulated using two SS‐PCD scans with each acquired with different tube potential and energy threshold settings, which also provided four energy‐bin datasets. The DS‐PCD approach was evaluated using both head and body phantoms. The radiation dose used in our phantom experiments was representative of routine clinical CT exams for head and body, which allowed us to evaluate the performance of DS‐PCD under a clinically encountered scenario. The results clearly demonstrated that DS‐PCD can improve the spectral separation and mass concentration quantification for I, Gd, and Bi. Because we used FBP reconstruction and least‐square based material decomposition method which are linear operators, the noise of the material decomposition results acquired with different dose levels can be predicted based on an inverse square root relationship between noise and dose.

Due to various technical limitations, the multi‐energy datasets acquired with SS‐PCD‐CT can suffer from imperfect PCD response. The imperfect PCD response causes a long tail in low energy‐bin energy spectra that spreads into high‐energy range, which decreases spectral separations between these datasets. As demonstrated in previous works,35, 42 energy separation between multi‐energy datasets affects image noise in the results of material decomposition. In this work, we demonstrated decreased image noise and improved quantification accuracy using DS‐PCD images. This can be explained by improved energy separation among the multiple energy‐bin datasets35, 42. The DS‐PCD reduces the high‐energy tails in low energy‐bin spectra by using a lower tube potential to acquire the lower energy‐bin datasets. The improvement in image quality (i.e., decreased image noise) and increased quantification accuracy was more significant for I and Gd compared to Bi. This is because the K‐edge of Gd is around 50 keV, and the spectral response is worst in the lower energy range. In contrast, Bi has a K‐edge around 90 keV, where the spectra are not as affected by the long inaccurate high‐energy tail.

The improved multi‐contrast imaging performance of the DS‐PCD approach can potentially benefit scenarios that require simultaneous imaging of multiple nanoparticle‐based contrast agents that have distinct K‐edges; the successful application of such novel contrasts will rely substantially on accurate material quantification. The DS‐PCD approach may also improve the performance of multi‐contrast, single phase imaging, which, if successful, may substantially reduce the radiation dose required to obtain clinically acceptable image quality compared to multi‐phase scanning.

Several studies have shown that the PCD‐CT have various advantages compared to the conventional EID‐CT, including improved iodine contrast, reduced electronic noise, more stable CT number in low dose realm, and dose‐efficient high‐resolution capability.6, 7, 8, 9, 10 The DS‐PCD system is expected to preserve these advantages as it is equipped with the same core PCD technology. These advantages can potentially be used to improve image quality under the same radiation dose, or to reduce patient dose while maintain the same image quality such as resolution and noise level. In addition, the DS‐PCD can be more dose efficient for multi‐energy applications as demonstrated here and in a previous work.37

In this work, we used filtered back projection reconstruction to generate the CT images and a least‐square based method without denoising to perform material decomposition. This was done to compare the underlying performance of the different approaches without potentially confounding the results by inclusion of other factors, such as iterative reconstruction. The material decomposition performance shown here most likely can be further improved if iterative reconstruction and material decomposition methods that include noise reduction techniques were to be used.43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53 Note that DS technology required image‐based decomposition methods, which were indeed used for this study.

Similar to the current DSDE‐CT systems using EID, a DS‐PCD system does not have co‐registered projection data between the two subsystems. It is therefore not straightforward how projection domain denoising using co‐registered projection data should be performed with DS‐PCD. However, image domain spectral denoising techniques can be applied to a DS system to reduce image noise, as have been implemented on commercial CT systems. Such algorithms have been tested in various clinical applications and been shown to be effective.54, 55, 56, 57 The loss of co‐registration between DSCT does not necessarily cause loss of spatial resolution, because the two subsystems are synchronized with each other during acquisition, and both of them have the same spatial resolution as a single‐source CT.

The performance of multi‐contrast material decomposition depends on object size; increased object size (i.e., abdomen vs head‐size object) can reduce multi‐contrast imaging performance. This is likely due to the hardening of the x‐ray beam as the object size increases, which shifts the mean photon energy in each energy bin toward higher energy values and reduces the differences in mean energies between different energy bins. Thus, we evaluated the DS‐PCD approach for both head and body phantoms of standard sizes. Imaging patients of different sizes, particularly for very small and very large patients, will require further investigation to optimize the tube potential and energy threshold settings, similar to what has been done for DECT using EID DSCT systems.42, 58, 59

Similar to studies with EID DECT using DS systems, an additional tin filter was used to further improve energy separation between data acquired using low and high tube potentials.35 However, other filtration materials, such as gold, Gd, or a combination of multiple materials, may also be used to improve the separation of different energy bins.60, 61 The choice of filtration materials, as well as their thicknesses, must be optimized based on the specific‐imaging task and the contrast materials of interest; such optimization might further improve material decomposition performance.

This study was performed based on a research whole‐body PCD‐CT system. This system reflects the performance of the only whole‐body PCD‐CT system that is currently available. As a research device, it has limitations that would not be present in a commercial system. For example, the chess mode, with its 50% dose efficiency, was a convenient method of providing simultaneous measurement of four energy thresholds at an early stage in the technology development, when detector pixels with only two energy thresholds were available for use in the project. The chess mode, with its inherent dose penalty, would not be used in a commercial system. With further technical advances in PCD technology, and further optimization of detector and system design based on knowledge obtained with the research system, considerable improvement in PCD‐CT performance is anticipated.

One limitation of this study is that cross scattering between two detector arrays when dual‐source data are simultaneously acquired was not considered in the phantom experiment. In reality, the cross scattering can reduce the energy separation between the low‐ and high‐energy subsystems, and therefore degrade decomposition performance. On the other hand, cross scattering correction techniques can be considered. On the current dual‐source CT systems, cross scattering also exists, and can usually be corrected to an acceptable level.62, 63 Similar methods could be used for cross scattering correction in the DS‐PCD system, although the difference between EID and PCD on the effect of cross scattering is an interesting area that worth further investigation.

The purpose of this preliminary study was to demonstrate the feasibility of performing multi‐contrast imaging on a DS PCD platform. To achieve this goal, we scanned contrast solutions with concentrations typically seen in literature.25 This setup allows us to use contrast material RMSE to compare the quantitative performance of DS‐PCD and SS‐PCD. Detection of low contrast agent concentration may require significant effort on developing appropriate postprocessing pipeline to achieve optimal performance, which may require the use of iterative reconstruction, additional image denoising, and material decomposition technique with noise reduction. Therefore, considerable efforts are required to properly establish the lowest concentration that can be resolved for a particular application of interest.

5. Conclusion

In this work, we explored the feasibility of using DS two energy‐bin PCD‐CT system design to perform simultaneous imaging of I, Gd, and Bi contrast materials, and compared its performance with an experimental whole‐body SS‐PCD system using chess acquisition mode with four energy bins. Phantom experiment results under an ideal condition without cross scattering demonstrated that DS‐PCD reduced material quantification RMSE and reduced noise in the material‐specific images compared to the SS‐PCD‐CT using chess mode, therefore showing the potential of DS‐PCD‐CT in simultaneous multi‐contrast imaging.

Conflict of Interest

CHM is the recipient of a research grant from Siemens Healthcare. The other authors have no relevant conflict of interest to disclose.

Acknowledgments

Research reported in this article was supported by the National Institutes of Health under award numbers R01 EB016966 and C06 RR018898. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Health.

References

  • 1. McCollough C, Leng S, Yu L, Fletcher JG. Dual‐and multi‐energy computed tomography: principles, technical approaches, and clinical applications. Radiology. 2015;276:637. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Bartlett DJ, Koo CW, Bartholmai BJ, et al. High‐resolution chest computed tomography imaging of the lungs: impact of 1024 matrix reconstruction and photon‐counting detector computed tomography. Invest Radiol. 2019;54:129–137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3. Marcus RP, Fletcher JG, Ferrero A, et al. Detection and characterization of renal stones by using photon‐counting‐based CT. Radiology. 2018;289:436–442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Pourmorteza A, Symons R, Sandfort V, et al. Abdominal imaging with contrast‐enhanced photon‐counting CT: first human experience. Radiology. 2016;279:239–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Zhou W, Bartlett DJ, Diehn FE, et al. Reduction of metal artifacts and improvement in dose efficiency using photon‐counting detector computed tomography and tin filtration. Invest Radiol. 2019;54:204–211. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Yu Z, Leng S, Kappler S, et al. Noise performance of low‐dose CT: comparison between an energy integrating detector and a photon counting detector using a whole‐body research photon counting CT scanner. J Med Imaging. 2016;3:043503. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Gutjahr R, Halaweish AF, Yu Z, et al. Human imaging with photon counting‐based computed tomography at clinical dose levels: contrast‐to‐noise ratio and cadaver studies. Invest Radiol. 2016;51:421–429. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8. Yu Z, Leng S, Jorgensen SM, et al. Evaluation of conventional imaging performance in a research whole‐body CT system with a photon‐counting detector array. Phys Med Biol. 2016;61:1572–1595. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Leng S, Rajendran K, Gong H, et al. 150‐μm spatial resolution using photon‐counting detector computed tomography technology: technical performance and first patient images. Invest Radiol. 2018;53:655–662. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10. Leng S, Zhou W, Yu Z, et al. Spectral performance of a whole‐body research photon counting detector CT: quantitative accuracy in derived image sets. Phys Med Biol. 2017;62:7216–7232. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11. Symons R, Krauss B, Sahbaee P, et al. Photon‐counting CT for simultaneous imaging of multiple contrast agents in the abdomen: An in vivo study. Med Phys. 2017;44:5120–5127. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Cooper DM, Chapman LD, Carter Y, et al. Three dimensional mapping of strontium in bone by dual energy K‐edge subtraction imaging. Phys Med Biol. 2012;57:5777–5786. [DOI] [PubMed] [Google Scholar]
  • 13. Feuerlein S, Roessl E, Proksa R, et al. Multienergy photon‐counting K‐edge imaging: potential for improved luminal depiction in vascular imaging. Radiology. 2008;249:1010–1016. [DOI] [PubMed] [Google Scholar]
  • 14. Muenzel D, Bar‐Ness D, Roessl E, et al. Spectral photon‐counting CT: initial experience with dual‐contrast agent K‐edge colonography. Radiology. 2017;283:723–728. [DOI] [PubMed] [Google Scholar]
  • 15. Riederer SJ, Mistretta CA. Selective iodine imaging using K‐edge energies in computerized x‐ray tomography. Med Phys. 1977;4:474–481. [DOI] [PubMed] [Google Scholar]
  • 16. Roessl E, Brendel B, Engel KJ, Schlomka JP, Thran A, Proksa R. Sensitivity of photon‐counting based K‐edge imaging in X‐ray computed tomography. IEEE Trans Med Imaging. 2011;30:1678–1690. [DOI] [PubMed] [Google Scholar]
  • 17. Roessl E, Proksa R. K‐edge imaging in x‐ray computed tomography using multi‐bin photon counting detectors. Phys Med Biol. 2007;52:4679–4696. [DOI] [PubMed] [Google Scholar]
  • 18. Schlomka JP, Roessl E, Dorscheid R, et al. Experimental feasibility of multi‐energy photon‐counting K‐edge imaging in pre‐clinical computed tomography. Phys Med Biol. 2008;53:4031–4047. [DOI] [PubMed] [Google Scholar]
  • 19. Muenzel D, Daerr H, Proksa R, et al. Simultaneous dual‐contrast multi‐phase liver imaging using spectral photon‐counting computed tomography: a proof‐of‐concept study. Eur Radiol Exp. 2017;1:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Symons R, Cork TE, Lakshmanan MN, et al. Dual‐contrast agent photon‐counting computed tomography of the heart: initial experience. Int J Cardiovasc Imaging. 2017;33:1253–1261. [DOI] [PubMed] [Google Scholar]
  • 21. Annapragada AV, Hoffman E, Divekar A, Karathanasis E, Ghaghada KB. High‐resolution CT vascular imaging using blood pool contrast agents. Methodist Debakey Cardiovasc J. 2012;8:18–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Cormode DP, Roessl E, Thran A, et al. Atherosclerotic plaque composition: analysis with multicolor CT and targeted gold nanoparticles. Radiology. 2010;256:774–782. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23. Ghaghada KB, Sato AF, Starosolski ZA, Berg J, Vail DM. Computed tomography imaging of solid tumors using a liposomal‐iodine contrast agent in companion dogs with naturally occurring cancer. PLoS ONE. 2016;11:e0152718. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Pan D, Schirra CO, Senpan A, et al. An early investigation of ytterbium nanocolloids for selective and quantitative "multicolor" spectral CT imaging. ACS Nano. 2012;6:3364–3370. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Dangelmaier J, Bar‐Ness D, Daerr H, et al. Experimental feasibility of spectral photon‐counting computed tomography with two contrast agents for the detection of endoleaks following endovascular aortic repair. Eur Radiol. 2018;28:3318–3325. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Huprich JE, Barlow JM, Hansel SL, Alexander JA, Fidler JL. Multiphase CT enterography evaluation of small‐bowel vascular lesions. AJR Am J Roentgenol. 2013;201:65–72. [DOI] [PubMed] [Google Scholar]
  • 27. Menon BK, d'Esterre CD, Qazi EM, et al. Multiphase CT angiography: a new tool for the imaging triage of patients with acute ischemic stroke. Radiology. 2015;275:510–520. [DOI] [PubMed] [Google Scholar]
  • 28. Nawfel RD, Judy PF, Schleipman AR, Silverman SG. Patient radiation dose at CT urography and conventional urography. Radiology. 2004;232:126–132. [DOI] [PubMed] [Google Scholar]
  • 29. Qu M, Ehman E, Fletcher JG, et al. Toward biphasic computed tomography (CT) enteric contrast: material classification of luminal bismuth and mural iodine in a small‐bowel phantom using dual‐energy CT. J Comput Assist Tomogr. 2012;36:554–559. [DOI] [PubMed] [Google Scholar]
  • 30. Cormode DP, Si‐Mohamed S, Bar‐Ness D, et al. Multicolor spectral photon‐counting computed tomography: in vivo dual contrast imaging with a high count rate scanner. Sci Rep. 2017;7:4784. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31. Kim JC, Anderson SE, Kaye W, et al. Charge sharing in common‐grid pixelated CdZnTe detectors. Nucl Instrum Methods Phys Res, Sect A. 2011;654:233–243. [Google Scholar]
  • 32. Koenig T, Zuber M, Hamann E, et al. How spectroscopic x‐ray imaging benefits from inter‐pixel communication. Phys Med Biol. 2014;59:6195. [DOI] [PubMed] [Google Scholar]
  • 33. Taguchi K, Iwanczyk JS. Vision 20/20: single photon counting x‐ray detectors in medical imaging. Med Phys. 2013;40:100901. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. Willemink MJ, Persson M, Pourmorteza A, Pelc NJ, Fleischmann D. Photon‐counting CT: technical principles and clinical prospects. Radiology. 2018;289:293–312. [DOI] [PubMed] [Google Scholar]
  • 35. Primak AN, Giraldo JC, Eusemann CD, et al. Dual‐source dual‐energy CT with additional tin filtration: dose and image quality evaluation in phantoms and in vivo. AJR Am J Roentgenol. 2010;195:1164–1174. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36. Faby S, Kuchenbecker S, Sawall S, et al. Performance of today's dual energy CT and future multi energy CT in virtual non‐contrast imaging and in iodine quantification: a simulation study. Med Phys. 2015;42:4349–4366. [DOI] [PubMed] [Google Scholar]
  • 37. Tao A, Huang R, Tao S, Michalak GJ, McCollough CH, Leng S. Dual‐source photon counting detector CT with a tin filter: a phantom study on iodine quantification performance. Phys Med Biol. 2019;64:115019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38. Li Z, Leng S, Yu Z, Kappler S, McCollough CH. Estimation of signal and noise for a whole‐body research photon‐counting CT system. J Med Imaging (Bellingham). 2017;4:023505. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39. Stierstorfer K. Modeling the frequency‐dependent detective quantum efficiency of photon‐counting x‐ray detectors. Med Phys. 2018;45:156–166. [DOI] [PubMed] [Google Scholar]
  • 40. Stierstorfer K, Rauscher A, Boese J, Bruder H, Schaller S, Flohr T. Weighted FBP–a simple approximate 3D FBP algorithm for multislice spiral CT with good dose usage for arbitrary pitch. Phys Med Biol. 2004;49:2209–2218. [DOI] [PubMed] [Google Scholar]
  • 41. Siewerdsen JH, Cunningham IA, Jaffray DA. A framework for noise‐power spectrum analysis of multidimensional images. Med Phys. 2002;29:2655–2671. [DOI] [PubMed] [Google Scholar]
  • 42. Primak AN, Ramirez Giraldo JC, Liu X, Yu L, McCollough CH. Improved dual‐energy material discrimination for dual‐source CT by means of additional spectral filtration. Med Phys. 2009;36:1359–1369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. Geyer LL, Schoepf UJ, Meinel FG, et al. State of the art: iterative CT reconstruction techniques. Radiology. 2015;276:339–357. [DOI] [PubMed] [Google Scholar]
  • 44. Korn A, Fenchel M, Bender B, et al. Iterative reconstruction in head CT: image quality of routine and low‐dose protocols in comparison with standard filtered back‐projection. AJNR Am J Neuroradiol. 2012;33:218–224. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45. Kalender WA, Klotz E, Kostaridou L. An algorithm for noise suppression in dual energy CT material density images. IEEE Trans Med Imaging. 1988;7:218–224. [DOI] [PubMed] [Google Scholar]
  • 46. Niu T, Dong X, Petrongolo M, Zhu L. Iterative image‐domain decomposition for dual‐energy CT. Med Phys. 2014;41:041901. [DOI] [PubMed] [Google Scholar]
  • 47. Tao S, Rajendran K, McCollough CH, Leng S. Material decomposition with prior knowledge aware iterative denoising (MD‐PKAID). Phys Med Biol. 2018;63:195003. [DOI] [PubMed] [Google Scholar]
  • 48. Cai C, Rodet T, Legoupil S, Mohammad‐Djafari A. A full‐spectral Bayesian reconstruction approach based on the material decomposition model applied in dual‐energy computed tomography. Med Phys. 2013;40:111916. [DOI] [PubMed] [Google Scholar]
  • 49. Long Y, Fessler JA. Multi‐material decomposition using statistical image reconstruction for spectral CT. IEEE Trans Med Imaging. 2014;33:1614–1626. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50. Mechlem K, Ehn S, Sellerer T, et al. Joint statistical iterative material image reconstruction for spectral computed tomography using a semi‐empirical forward model. IEEE Trans Med Imaging. 2018;37:68–80. [DOI] [PubMed] [Google Scholar]
  • 51. Mory C, Sixou B, Si‐Mohamed S, Boussel L, Rit S. Comparison of five one‐step reconstruction algorithms for spectral CT. Phys Med Biol. 2018;63:235001. [DOI] [PubMed] [Google Scholar]
  • 52. Weidinger T, Buzug TM, Flohr T, Kappler S, Stierstorfer K. Polychromatic iterative statistical material image reconstruction for photon‐counting computed tomography. Int J Biomed Imaging. 2016;2016:5871604. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53. Foygel Barber R, Sidky EY, Gilat Schmidt T, Pan X. An algorithm for constrained one‐step inversion of spectral CT data. Phys Med Biol. 2016;61:3784–3818. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54. Fulwadhva UP, Wortman JR, Sodickson AD. Use of dual‐energy CT and iodine maps in evaluation of bowel disease. Radiographics. 2016;36:393–406. [DOI] [PubMed] [Google Scholar]
  • 55. Pelgrim GJ, van Hamersvelt RW, Willemink MJ, et al. Accuracy of iodine quantification using dual energy CT in latest generation dual source and dual layer CT. Eur Radiol. 2017;27:3904–3912. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56. Patino M, Prochowski A, Agrawal MD, et al. Material separation using dual‐energy CT: current and emerging applications. Radiographics. 2016;36:1087–1105. [DOI] [PubMed] [Google Scholar]
  • 57. Grant KL, Flohr TG, Krauss B, Sedlmair M, Thomas C, Schmidt B. Assessment of an advanced image‐based technique to calculate virtual monoenergetic computed tomographic images from a dual‐energy examination to improve contrast‐to‐noise ratio in examinations using iodinated contrast media. Invest Radiol. 2014;49:586–592. [DOI] [PubMed] [Google Scholar]
  • 58. Michalak G, Grimes J, Fletcher J, et al. Selection of optimal tube potential settings for dual‐energy CT virtual mono‐energetic imaging of iodine in the abdomen. Abdom Radiol (NY). 2017;42:2289–2296. [DOI] [PubMed] [Google Scholar]
  • 59. Wang AS, Pelc NJ. Sufficient statistics as a generalization of binning in spectral X‐ray imaging. IEEE Trans Med Imaging. 2011;30:84–93. [DOI] [PubMed] [Google Scholar]
  • 60. Shikhaliev PM. Photon counting spectral CT: improved material decomposition with K‐edge‐filtered x‐rays. Phys Med Biol. 2012;57:1595–1615. [DOI] [PubMed] [Google Scholar]
  • 61. Yu L, Ren L, Li Z, Leng S, McCollough CH. Dual‐source multienergy CT with triple or quadruple x‐ray beams. J Med Imaging (Bellingham). 2018;5:033502. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62. Petersilka M, Stierstorfer K, Bruder H, Flohr T. Strategies for scatter correction in dual source CT. Med Phys. 2010;37:5971–5992. [DOI] [PubMed] [Google Scholar]
  • 63. Mobberley SD, Fuld MK, Sieren JP, Primak AN, Hoffman EA. Scatter correction associated with dedicated dual‐source CT hardware improves accuracy of lung air measures. Acad Radiol. 2013;20:1334–1343. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Medical Physics are provided here courtesy of American Association of Physicists in Medicine

RESOURCES