Image Quality Assessment for Clinical Cadmium Telluride-Based Photon-Counting Computed Tomography Detector in Cadaveric Wrist Imaging : Investigative Radiology

Secondary Logo

Journal Logo

Original Articles

Image Quality Assessment for Clinical Cadmium Telluride-Based Photon-Counting Computed Tomography Detector in Cadaveric Wrist Imaging

Grunz, Jan-Peter MD; Huflage, Henner MD; Heidenreich, Julius Frederik MD; Ergün, Süleyman MD; Petersilka, Martin PhD; Allmendinger, Thomas PhD; Bley, Thorsten Alexander MD; Petritsch, Bernhard MD

Author Information
doi: 10.1097/RLI.0000000000000789
  • Open

Abstract

Fractures of the wrist are among the most common injuries in any emergency department. Depending on the impact, trauma mechanism, and bone tissue of the patient, wrist fractures can manifest in a multitude of patterns, ranging from subtle fissures in the distal radius to severely dislocated multifragment situations with concomitant injuries of the carpus.1,2 From a radiological perspective, fracture suspicion is primarily evaluated on radiograms. However, radiographically occult injuries or assessment of joint surface affliction oftentimes benefit from additional computed tomography (CT) imaging.3,4 With fast scan times through helical acquisition and ubiquitous availability, energy-integrating multidetector CT systems are the current clinical standard for 3D imaging in trauma. Because of limited spatial resolution, as well as susceptibility to beam hardening artifacts and partial volume effects, wrist trauma evaluation remains a challenging task for radiologists with the current generation of energy-integrating CT detectors, though. Especially articular affliction and subtle trabecular fractures of the distal forearm and carpal bones can be difficult to evaluate.5 Detector models with smaller cell size may have the potential to address these limitations; however, in energy-integrating builds, geometric efficiency remains hampered by constructional restrictions, for example, the necessity to include optical septa between detector elements.6 Furthermore, down-weighting of low-energy photons and noise susceptibility limit the potential for radiation dose reduction.7,8 The photon-counting detector (PCD) technology represents a promising approach to overcome these limitations. In contrast to energy-integrating detector (EID) models, which need to perform photon conversion into light to generate images, PCDs are composed of semiconductors (eg, cadmium telluride) that use incoming x-rays to create electron-hole pairs.9 Although conventional builds depend on optical separation of detector elements, which decreases dose efficiency, detector cells in PCDs are marked by an electric field between the common cathode and pixelated anodes, hence eliminating the need for further separation layers.10 Thus far, preclinical PCDs have primarily been evaluated in phantom studies, for example, for vascular and stent lumen assessment, in anthropomorphic body imaging, or for ex vivo differentiation of crystal depositions in joint calcification.11–14 Although an earlier prototype version of the detector used in this study has been tested for in vitro temporal bone imaging,6,15 image analyses for both the current clinical detector model and the appendicular skeleton are lacking. We hypothesized that the clinical PCD in this study can provide substantial advantages for the assessment of bone microarchitecture over CT systems with conventional EID builds. Therefore, this work aims to provide a subjective and semiquantitative image quality comparison between the 2 detector technologies for cadaveric wrist scans.

MATERIALS AND METHODS

Cadaveric Specimens and Wrist Positioning

Eight formalin-fixed cadaveric wrists of 4 body donors were obtained from the anatomical institute of our university. All wrists were examined with a clinical PCD-CT system (SOMATOM CountPlus; Siemens Healthcare GmbH, Forchheim, Germany) and a third-generation dual-source CT scanner (SOMATOM Force, Siemens). The study was approved by the ethics committee of our medical faculty and conducted in accordance with institutional laws and regulations. Body donors consented to the use of their cadavers for study and research purposes during their lifetime. Irrespective of scanner, examinations were performed with the cadaveric specimens in prone position and 1 arm extended above the head parallel to the z axis of the system. Wrists were positioned within the isocenter of the respective gantry to provide the least possible image noise.16,17 A flowchart summarizing the study setup is presented in Figure 1.

F1
FIGURE 1:
Flowchart for visualization of study design.

Technical Specifications and Scan Protocols

The photon-counting CT system used in this study is derived from an earlier scanner prototype that was based on the architecture of a second-generation dual-source CT.6,18 It is equipped with a clinical PCD that uses cadmium telluride semiconductors for image generation and an x-ray tube (Vectron, Siemens) with focal spot width × length of 0.7 × 0.9 mm in ultra-high-resolution (UHR-PCD) and 1.0 × 1.4 mm in standard-resolution (SR-PCD) mode. The detector consists of subpixels measuring 0.275 × 0.322 mm that can be read out separately in UHR-PCD or after 2 × 2 binning to form “macropixels” in SR-PCD mode. The maximum cross-axial field of view is 500 mm, and 4 energy-thresholds can be discerned per subpixel. Z-coverage at the isocenter in standard-resolution mode and ultra-high-resolution mode is 57.6 mm (144 × 0.4 mm) and 24.0 mm (120 × 0.2 mm), respectively.

For the commercially available EID-CT scanner, we implemented 3 scan protocols with varying tube currents and fixed reference tube voltage of 120 kVp in single-energy mode: a dedicated low-dose protocol with 25 mAs (volume CT dose index = 1.50 mGy), a clinical standard protocol with 100 mAs (5.80 mGy), and a full-dose protocol with 150 mAs for optimal image quality (8.67 mGy). The EID-CT scanner depends on a mechanical comb filter in front of the detector (with coverage limited to 32 rows) to facilitate ultra-high-resolution scans. Using the slice doubling effect of periodic focal spot motion in z-direction, all EID-CT scans were performed with the maximum UHR collimation of 2 × 32 × 0.6 mm, gantry rotation time of 1000 milliseconds, and pitch factor of 0.8. With equal tube voltage of 120 kVp, the scan protocols for the photon-counting CT system were chosen to match the volume CT dose indices of the conventional CT studies, resulting in tube current-time products of 17 mAs (low-dose), 70 mAs (standard-dose), and 107 mAs (full-dose). Standard-resolution PCD acquisitions were performed with helical pitch factor of 0.6 and rotation time of 500 milliseconds, whereas UHR-PCD scans were acquired with pitch of 1.5 and rotation time of 1000 milliseconds. Acquisition settings were chosen with respect to the physical limitations of each scan mode to provide reasonable comparison to EID-CT scans.

Image Reconstruction Parameters

A fourth-generation iterative reconstruction algorithm was used for reformatting of raw data from the PCD (QIR, Siemens). All data sets were reconstructed with strength level 3 of 4 using spectral information to offset potential cone-beam and beam-hardening artifacts. In addition, the iterative reconstruction algorithm performs denoising based on a 20-keV energy threshold.7 For scans with the EID, a commercially available third-generation iterative reconstruction algorithm (ADMIRE, Siemens) was applied with comparable strength setting. As bone analysis was the primary focus of this study, the kernels used for photon-counting CT reconstructions (UHR-PCD, Br76; SR-PCD, Br68; both Siemens) were chosen as close as technically possible to the modulation transfer function (MTF) of a sharp bone kernel used in conventional EID-CT imaging (Ur77, Siemens). It must be noted, however, that the kernel selected for reformatting of UHR-PCD data operates below the spatial resolution maximum of the cadmium telluride-based detector.

To achieve identical in-plane resolution, all raw data from both systems were reformatted with a field of view of 100 mm and image matrix of 768 × 768 pixels. Aiming to match the in-plane resolution of each scanner and acquisition mode, slice thickness and increment were chosen as small as technically possible (UHR-PCD: thickness, 0.2 mm; increment, 0.1 mm; SR-PCD: thickness, 0.4 mm; increment, 0.3 mm; EID-CT: thickness, 0.6 mm; increment, 0.4 mm). Preset window settings were 2800/800 HU (window width/center) for optimized bone evaluation, but observers could adjust the contrast to their requirements.

Subjective Image Quality Assessment

Reconstructed images were independently analyzed by 3 radiologists with 4 (J.F.H.), 4 (J.P.G.), and 5 (H.H.) years of skeletal imaging experience in randomized order and blinded fashion. Image analysis was conducted in standardized conditions with dedicated picture archiving and communication software (Merlin; Phönix-PACS, Freiburg, Germany) and certified diagnostic monitors (RadiForce RX660; EIZO, Hakusan, Japan). After the initial review of all studies, observers were tasked to state which images were suitable for fracture assessment. In addition, subjective evaluation of image quality was performed based on assessability of cortical bone, cancellous bone, and soft tissue using a 7-point scale (excellent, 7; very good, 6; good, 5; satisfactory, 4; fair, 3; poor, 2; very poor, 1).

Objective Image Quality Assessment

Computed tomography number measurements were performed by a radiologist with 4 years of skeletal imaging experience (J.P.G.) using specific 3D reading software (syngo.via version VB40B, Siemens). Avoiding extensive artifacts, identical circular regions of interest (ROIs) were placed in consistent locations within the cancellous bone of the hamate and adjacent subcutaneous fat tissue on 3 consecutive axial CT slices (Fig. 2). Mean signal attenuation and standard deviation were recorded in Hounsfield units for each ROI and averaged across slices to ensure measurement consistency. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were computed with noise defined as the standard deviation of CT numbers within fat tissue, because subcutaneous fat provides a more homogenous texture compared with cancellous bone.19

F2
FIGURE 2:
Depiction of CT number measurements within regions of interest placed in the hamate bone and the adjacent subcutaneous fat tissue on 3 consecutive ultra-high-resolution photon-counting CT slices.

Statistics

A dedicated statistical software was used for data analysis (SPSS Statistics Version 27.0 for Mac; IBM, Armonk, NY). Normal distribution of continuous variables was analyzed with Kolmogorov-Smirnov tests. For categorical variables, we report absolute numbers, percentages, and median values, whereas means ± standard deviations are presented for normally distributed continuous variables. The intraclass correlation coefficient was calculated to assess interrater reliability for overall image quality based on a 2-way random-effects model and absolute agreement with interpretation of reliability following Koo and Li (excellent, 1.00–0.90; good, 0.90–0.75; moderate, 0.75–0.50; poor, <0.50).20 Paired nonparametric variables were compared with Wilcoxon signed rank tests. Statistical significance was assumed for P values less than 0.05.

RESULTS

Using 3 dose protocols each with the conventional energy-integrating CT and both scan modes of the photon-counting CT, we performed a total of 72 wrist examinations in this study. Representative axial images of the distal radius with each combination are presented in Figure 3.

F3
FIGURE 3:
Representative CT slices at the level of the distal radius for all 9 combinations of dose protocol and scanner/scan mode. Left column, conventional energy-integrating CT (A1, low-dose; A2, standard-dose; A3, full-dose). Middle column, standard-resolution photon-counting CT (B1, low-dose; B2, standard-dose; B3, full-dose). Right column, ultra-high-resolution photon-counting CT (C1, low-dose; C2, standard-dose; C3, full-dose).

Subjective Image Quality Assessment

Irrespective of scanner, the image quality of all full-dose and standard-dose studies was deemed sufficient for fracture evaluation. For low-dose scans, readers considered 50% of EID-CT, 79% of SR-PCD, and 96% of UHR-PCD suitable for clinical use. Between dose equivalent scan protocols, subjective image quality assessment provided superior results for UHR-PCD in comparison with SR-PCD (all P's < 0.001) and EID-CT (all P's ≤ 0.002). Performance of standard-dose UHR-PCD was also rated preferably compared with full-dose SR-PCD (P = 0.016) and EID-CT scans (P = 0.040). No distinction was found between the image quality of low-dose UHR-PCD and standard-dose studies with both SR-PCD (P = 0.108) and EID-CT (P = 0.470). Comparison of SR-PCD and EID-CT provided no considerable difference for full-dose (P = 0.248) and standard-dose examinations (P = 0.509), whereas SR-PCD received better image quality ratings in low-dose scans (P < 0.001). Subjective image quality ratings are summarized in Table 1. Indicated by the single-measure intraclass correlation coefficient of 0.876 (95% confidence interval, 0.744–0.925; P < 0.001), interrater reliability was good throughout this study.

TABLE 1 - Subjective Evaluation of Image Quality
Scale EID-CT SR-PCD UHR-PCD
Value LD SD FD LD SD FD LD SD FD
Image quality 7 0 2 (8.3) 3 (12.5) 0 0 1 (4.2) 0 11 (45.8) 15 (62.5)
6 0 6 (25.0) 14 (58.3) 0 11 (45.8) 15 (62.5) 9 (37.5) 8 (33.3) 7 (29.2)
5 2 (8.3) 11 (45.8) 5 (20.8) 9 (37.5) 8 (33.3) 6 (25.0) 5 (20.8) 4 (16.7) 2 (8.3)
4 2 (8.3) 3 (12.5) 1 (4.2) 7 (29.2) 3 (12.5) 2 (8.3) 9 (37.5) 0 0
3 9 (37.5) 0 1 (4.2) 4 (16.7) 2 (8.3) 0 0 1 (4.2) 0
2 6 (25.0) 2 (8.3) 0 3 (12.5) 0 0 1 (4.2) 0 0
1 5 (20.8) 0 0 1 (4.2) 0 0 0 0 0
Median 3.0 5.0 6.0 4.0 5.0 6.0 5.0 6.0 7.0
Combined image quality ratings of the 3 observers for photon-counting and energy-integrating CT scans (1 = very poor; 2 = poor; 3 = fair; 4 = satisfactory; 5 = good; 6 = very good; 7 = excellent). Results are reported as frequencies (percentages) and median values.
EID-CT, energy-integrating CT; SR-PCD, standard-resolution photon-counting CT; UHR-PCD, ultra-high-resolution photon-counting CT; LD, low-dose protocol (CTDIvol = 1.50 mGy); SD, standard-dose protocol (CTDIvol = 5.80 mGy); FD, full-dose protocol (CTDIvol = 8.67 mGy).

Semiquantitative Image Quality Assessment

Between dose equivalent studies, SNR was considerably lower for EID-CT compared with dose equivalent SR-PCD and UHR-PCD (all P's < 0.001). Similarly, CNR values were superior with both operating modes of the photon-counting CT (P < 0.001). For detailed visualization of image noise, signal intensity, and contrast, Figure 4 contains head-to-head comparison of clinical dose images depicting an arthritic carpometacarpal joint. No significant difference was observed between standard-dose UHR-PCD (SNR/CNR, 4.49/4.70) and full-dose EID-CT (SNR/CNR, 4.41/4.64; P = 0.932/0.977), whereas standard-dose SR-PCD yielded higher SNR (5.22; P = 0.002) and CNR values (5.46; P = 0.002) than full-dose EID-CT scans. Direct comparison between the 2 scan modes of the clinical photon-counting CT resulted in favorable SNR and CNR for the SR-PCD on all dose levels (P < 0.001). The SNR and CNR values are summarized in Table 2.

F4
FIGURE 4:
Representative CT slices of an arthritic carpometacarpal joint with the clinical standard-dose protocol for visualization of the respective image quality of each scanner/scan mode. A, Conventional energy-integrating CT. B, Standard-resolution photon-counting CT. C, Ultra-high-resolution photon-counting CT.
TABLE 2 - Signal-to-Noise Ratio and Contrast-to-Noise Ratio Values
EID-CT SR-PCD UHR-PCD
Median SNR (range)
 LD 2.36 (0.24) 3.44 (0.31) 3.13 (0.41)
 SD 3.72 (0.65) 5.22 (0.83) 4.49 (0.68)
 FD 4.41 (0.32) 6.10 (0.77) 5.42 (0.36)
Median CNR (range)
 LD 2.52 (0.26) 3.59 (0.29) 3.38 (0.37)
 SD 3.86 (0.73) 5.46 (0.98) 4.70 (0.68)
 FD 4.64 (0.44) 6.35 (0.71) 5.63 (0.43)
Calculation of SNR and CNR was performed after ROI-based measurement of signal attenuation in bone and fat tissue. Image noise was defined as the standard deviation of CT numbers within fat tissue. Results are reported as median values and interquartile ranges.
EID-CT, energy-integrating CT; SR-PCD, standard-resolution photon-counting CT; UHR-PCD, ultra-high-resolution photon-counting CT; LD, low-dose protocol (CTDIvol = 1.50 mGy); SD, standard-dose protocol (CTDIvol = 5.80 mGy); FD, full-dose protocol (CTDIvol = 8.67 mGy).

DISCUSSION

For this study, we compared the performance of a CT system equipped with a clinical cadmium telluride-based PCD with state-of-the-art dual-source CT with EID technology. In dose-matched examinations of 8 cadaveric wrists, observer analysis favored the photon-counting CT's ultra-high-resolution scan mode, whereas semiquantitative image quality assessment resulted in higher SNR and CNR for both PCD operating modes over the commercially available EID system. To the authors' best knowledge, this study represents the first application of a clinical PCD to dedicated imaging of the wrist.

Earlier studies with preclinical PCD prototypes were able to show that noise reduction of approximately 30% to 40% can be achieved when optical separation is not required.21,22 Concordant with literature, SNR and CNR were considerably higher for wrist scans with the PCD in this study, enabling dose reduction of up to 75% compared with EID-CT without significant decrease in SNR or CNR. Consisting of semiconductors such as cadmium telluride, PCDs have the ability to directly convert incoming x-ray beams into voltage pulses, which are counted if they exceed a certain threshold. As count rates are not affected by low-level electronic noise below these predefined thresholds, overall image noise is considerably less pronounced even in low-dose examinations.23 Furthermore, unlike in EID builds, contribution of x-ray photons to the signal measurement is independent of their actual energy difference to the threshold.24 This may explain the superior SNR and CNR values of low-dose images in comparison to conventional CT studies with higher-dose levels.

Although noise reduction was most pronounced with the standard-resolution scan mode of the PCD-CT, subjective image evaluation yielded the best results for the ultra-high-resolution scan mode. This finding can most likely be attributed to the more detailed visualization of bone microarchitecture facilitated by the smaller focal spot and unbinned readout of the detector's subpixel structure. Whereas the standard-resolution mode performs detector readout after 2 × 2 binning, resulting in “macropixels” with edge length comparable to the EID build used in this study, the ultra-high-resolution scans achieve much sharper differentiation of trabecula and bone marrow. Reconstruction kernels for photon-counting CT scans were selected to match the convolution kernel that facilitates optimal bone visualization with the energy-integrating system (Ur77; ρ10 = 22.0 lp/cm, ρ50 = 16.5 lp/cm). However, the realized image sharpness of these kernels (UHR-PCD: Br76; ρ10 = 21.0 lp/cm, ρ50 = 16.5 lp/cm and SR-PCD: Br68; ρ10 = 14.4 lp/cm, ρ50 = 11.8 lp/cm) is far lower than the system's raw MTF in the UHR-PCD mode without a defined kernel. It must be noted that the sharpness characterized by the Br68 kernel represents the technical maximum for the standard-resolution photon-counting CT scan mode, although it operates with a lower MTF than the other reconstruction kernels in this study. This is due to the fact that the amount of detector channels (in-plane: 1376 channels) in combination with the focal spot size limits the achievable sharpness in this mode. As shown before, data acquisition with small pixel size and subsequent reconstruction with an MTF below the detector resolution limit facilitates noise reduction over images with larger pixels and equal reconstruction MTF.25,26 However, despite superior denoising, the resulting loss of sharpness might have been the reason for the similar image quality ratings of standard-resolution photon-counting and conventional CT. Further analysis of image reformatting with other convolution kernels is necessary to assess whether the PCD can realize sharper reformatting in ultra-high-resolution mode without extensive noise increase.

Limiting the interpretation of results, only 8 cadaveric wrists from 4 body donors were scanned in this in vitro study. Cadaveric specimens were chosen without knowledge of age, bone density, or duration of formalin fixation. Therefore, the extent of bone demineralization was heterogenous among the examined wrists and could have influenced image quality ratings.27 As this work was limited to uninjured cadaveric wrists, further studies are essential to evaluate the added value of the photon-counting technology for diagnostic accuracy in trauma imaging. Although the authors expect an increased sensitivity for fracture diagnosis, misinterpretation of the superiorly visible bone channels as fracture lines may negatively affect diagnostic specificity due to more false-positive cases. Besides, the effect of wrist motion and off-center positioning should be assessed in future patient studies to provide a more realistic comparison between scanners. Although observers were blinded to any information on image acquisition and reconstruction, they might have become familiar with certain image characteristics of each scan mode over the course of their evaluation. Without a defined scan or reconstruction protocol for photon-counting CT wrist examinations, we adopted acquisition parameters and selected reconstruction kernels in this study based on conventional CT scans in clinical routine, hence potentially underestimating the performance of the photon-counting scanner. Finally, to ensure comparability with the EID system, image data of the photon-counting CT were reconstructed with a single minimum threshold at 20 keV instead of separating the energy spectrum into different energy bins for virtual monoenergetic reconstructions.

CONCLUSIONS

Superior visualization of fine anatomy is feasible with the clinical photon-counting CT system in cadaveric wrist scans. The ultra-high-resolution scan mode suggests potential for considerable dose reduction over energy-integrating dual-source CT.

REFERENCES

1. Court-Brown CM, Caesar B. Epidemiology of adult fractures: a review. Injury. 2006;37:691–697.
2. Nellans KW, Kowalski E, Chung KC. The epidemiology of distal radius fractures. Hand Clin. 2012;28:113–125.
3. Torabi M, Lenchik L, Beaman FD, et al. ACR Appropriateness Criteria® acute hand and wrist trauma. J Am Coll Radiol. 2019;16:S7–S17.
4. Grunz JP, Pennig L, Fieber T, et al. Twin robotic x-ray system in small bone and joint trauma: impact of cone-beam computed tomography on treatment decisions. Eur Radiol. 2020.
5. Müller FC, Gosvig KK, Børgesen H, et al. Dual-energy CT for suspected radiographically negative wrist fractures: a prospective diagnostic test accuracy study. Radiology. 2020;296:596–602.
6. Zhou W, Lane JI, Carlson ML, et al. Comparison of a photon-counting-detector CT with an energy-integrating-detector CT for temporal bone imaging: a cadaveric study. Am J Neuroradiol. 2018;39:1733–1738.
7. Flohr T, Petersilka M, Henning A, et al. Photon-counting CT review. Phys Med. 2020;79:126–136.
8. McCollough CH, Boedeker K, Cody D, et al. Principles and applications of multienergy CT: Report of AAPM Task Group 291. Med Phys. 2020;47:e881–e912.
9. Wildberger JE, Prokop M. Hounsfield's Legacy. Invest Radiol. 2020;55:556–558.
10. Willemink MJ, Persson M, Pourmorteza A, et al. Photon-counting CT: technical principles and clinical prospects. Radiology. 2018;289:293–312.
11. Von Spiczak J, Mannil M, Peters B, et al. Photon counting computed tomography with dedicated sharp convolution kernels: tapping the potential of a new technology for stent imaging. Invest Radiol. 2018;53:486–494.
12. Sartoretti T, Eberhard M, Rüschoff JH, et al. Photon-counting CT with tungsten as contrast medium: experimental evidence of vessel lumen and plaque visualization. Atherosclerosis. 2020;310:11–16.
13. Klein L, Dorn S, Amato C, et al. Effects of detector sampling on noise reduction in clinical photon-counting whole-body computed tomography. Invest Radiol. 2020;55:111–119.
14. Huber FA, Becce F, Gkoumas S, et al. Differentiation of crystals associated with arthropathies by spectral photon-counting radiography: a proof-of-concept study. Invest Radiol. 2021;56:147–152.
15. Rajendran K, Voss BA, Zhou W, et al. Dose reduction for sinus and temporal bone imaging using photon-counting detector CT with an additional tin filter. Invest Radiol. 2020;55:91–100.
16. Kaasalainen T, Palmu K, Reijonen V, et al. Effect of patient centering on patient dose and image noise in chest CT. Am J Roentgenol. 2014;203:123–130.
17. Alkadhi H, Euler A. The future of computed tomography: personalized, functional, and precise. Invest Radiol. 2020;55:545–555.
18. Symons R, Reich DS, Bagheri M, et al. Photon-counting computed tomography for vascular imaging of the head and neck: first in vivo human results. Invest Radiol. 2018;53:135–142.
19. Yel I, Booz C, Albrecht MH, et al. Optimization of image quality and radiation dose using different cone-beam CT exposure parameters. Eur J Radiol. 2019;116:68–75.
20. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. 2016;15:155–163.
21. Leng S, Yu Z, Halaweish A, et al. Dose-efficient ultrahigh-resolution scan mode using a photon counting detector computed tomography system. J Med Imaging. 2016;3:043504.
22. 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.
23. Bornefalk H, Danielsson M. Photon-counting spectral computed tomography using silicon strip detectors: a feasibility study. Phys Med Biol. 2010;55:1999–2022.
24. Flohr T, Ulzheimer S, Petersilka M, et al. Basic principles and clinical potential of photon-counting detector CT. Chin J Acad Radiol. 2020;3:19–34.
25. 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.
26. Pourmorteza A, Symons R, Henning A, et al. Dose efficiency of quarter-millimeter photon-counting computed tomography: first-in-human results. Invest Radiol. 2018;53:365–372.
27. Fonseca AA, Cherubini K, Veeck EB, et al. Effect of 10% formalin on radiographic optical density of bone specimens. Dentomaxillofac Radiol. 2008;37:137–141.
Keywords:

photon-counting; tomography; x-ray computed; wrist; radiation dosage; cancellous bone

Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc.