Feasibility of Dose-reduced Chest CT with Photon-counting Detectors: Initial Results in Humans

This feasibility study demonstrated good diagnostic quality, noise power spectrum, and lung nodule contrast-to-noise ratio with dose-reduced photon-counting detector chest CT compared with those attained with conventional energy-integrating detector CT.

Abstract

Purpose

To investigate whether photon-counting detector (PCD) technology can improve dose-reduced chest computed tomography (CT) image quality compared with that attained with conventional energy-integrating detector (EID) technology in vivo.

Materials and Methods

This was a HIPAA-compliant institutional review board-approved study, with informed consent from patients. Dose-reduced spiral unenhanced lung EID and PCD CT examinations were performed in 30 asymptomatic volunteers in accordance with manufacturer-recommended guidelines for CT lung cancer screening (120-kVp tube voltage, 20-mAs reference tube current–time product for both detectors). Quantitative analysis of images included measurement of mean attenuation, noise power spectrum (NPS), and lung nodule contrast-to-noise ratio (CNR). Images were qualitatively analyzed by three radiologists blinded to detector type. Reproducibility was assessed with the intraclass correlation coefficient (ICC). McNemar, paired t, and Wilcoxon signed-rank tests were used to compare image quality.

Results

Thirty study subjects were evaluated (mean age, 55.0 years ± 8.7 [standard deviation]; 14 men). Of these patients, 10 had a normal body mass index (BMI) (BMI range, 18.5–24.9 kg/m²; group 1), 10 were overweight (BMI range, 25.0–29.9 kg/m²; group 2), and 10 were obese (BMI ≥30.0 kg/m², group 3). PCD diagnostic quality was higher than EID diagnostic quality (P = .016, P = .016, and P = .013 for readers 1, 2, and 3, respectively), with significantly better NPS and image quality scores for lung, soft tissue, and bone and with fewer beam-hardening artifacts (all P < .001). Image noise was significantly lower for PCD images in all BMI groups (P < .001 for groups 1 and 3, P < .01 for group 2), with higher CNR for lung nodule detection (12.1 ± 1.7 vs 10.0 ± 1.8, P < .001). Inter- and intrareader reproducibility were good (all ICC > 0.800).

Conclusion

Initial human experience with dose-reduced PCD chest CT demonstrated lower image noise compared with conventional EID CT, with better diagnostic quality and lung nodule CNR.

^© RSNA, 2017

Online supplemental material is available for this article.

Introduction

Photon-counting detectors (PCDs) use semiconductor materials, such as cadmium telluride, to directly convert each x-ray photon into an electric pulse. The amplitude of the pulses is proportional to the energy of the individual incident photons. High-speed application-specific integrated circuits embedded in the PCD count the number of pulses (ie, detected photons) above set energy thresholds (1–4). This effectively suppresses electronic noise in the photon counts, which may lead to substantial image noise reduction.

Conventional energy-integrating detectors (EIDs) combine the effects of multiple x-ray photons into an intensity value; hence, the contribution of low-energy photons is smaller than the relative contribution of high-energy photons. In contrast, PCDs measure the number and energy of incident x-ray photons separately. Thus, photons of different energies contribute equally to the photon counts. Hence, the weight of low-energy photons, which provide more soft-tissue contrast, is improved in PCDs (4). Spectral information of photon-counting computed tomography (CT) has been successfully used to separate up to three mixed contrast agents in vivo (5,6). However, even without using spectral information, PCDs may reduce beam-hardening artifacts and improve contrast-to-noise ratios (CNRs) due to the suppression of electronic noise and improved soft-tissue contrast. PCDs may be of particular interest for dose-reduced CT examinations, such as lung cancer screening CT, in which image noise and beam-hardening effects of conventional EIDs become prominent, imposing a lower limit on radiation dose reduction (7,8).

The suppression of electronic noise with PCD CT leads to better attenuation stability and lower image noise; this concept has been extensively evaluated in a specialized lung phantom for chest CT (9,10). PCD CT demonstrated up to 11.8% less image noise and 11.4% higher CNR in low-contrast lesion detectability (eg, ground-glass nodules and emphysema) for multiple combinations of tube current and tube voltage settings. A 10% reduction in image noise would translate to a potential 19% reduction in radiation dose without compromising diagnostic image quality. Given the common clinical indication of dose-reduced chest CT for lung cancer screening, this may result in major radiation dose reductions on a population level (11). The purpose of this study was to investigate whether PCD technology can improve dose-reduced chest CT image quality compared with that of conventional EID technology in vivo.

Materials and Methods

This study was supported by a collaborative research agreement with Siemens Medical Systems (Forchheim, Germany). Authors who are not employees of or consultants for Siemens Healthcare controlled the inclusion of any data and information that might have presented a conflict of interest for authors who are Siemens Healthcare employees.

Study Population

Thirty asymptomatic volunteers older than 45 years (14 men) were prospectively enrolled at the National Institutes of Health Clinical Center in this Health Insurance Portability and Accountability Act–compliant institutional review board–approved study with informed consent. Mean age for all volunteers was 55 years (range, 45–75 years); mean age for men was 55 years (age range, 45–72 years), and mean age for women was 55 years (age range, 45–75 years). There was no significant difference in age between men and women (P = .88, Student t test). Exclusion criteria included pregnancy, known or possible genetic disposition to radiation-induced cancer, and CT scanning within the previous 12 months. No subjects were excluded from analysis after enrollment.

PCD CT System

The whole-body prototype PCD CT system has been previously described (12,13). In brief, the PCD CT system is based on a dual-source CT scanner (Somatom Definition Flash; Siemens Healthcare, Forchheim, Germany) with two independent x-ray sources at 95° offset in which one of the conventional EID systems is replaced with a cadmium telluride PCD. This current prototype set-up with EID and a PCD subsystem, identical x-ray tubes, and beam-shaping filters provides a convenient platform for comparison between the two detector technologies. It should be noted that, if commercialized, the final product will most likely include only PCDs. The fields of view for EID and PCD are 50 cm and 27.5 cm, respectively, with collimation × pixel pitches of 64 × 0.6 mm and 32 × 0.5 mm, respectively, at the isocenter. The smaller field of view for the PCD is not a limitation of photon-counting technology but merely a design choice for this prototype scanner. In this study, we set the PCD energy thresholds at 25 keV and 50 keV. We used PCD images reconstructed from all detected photons with energies greater than 25 keV to compare both detectors. The high 50-keV threshold was chosen arbitrarily and was not used in this study, as dividing the limited number of detected photons in these dose-reduced scans would result in photon-starved energy bins (4). The goal of this study was to focus on the nonspectral advantages of photon-counting technology, such as reduced sensitivity to electronic noise and beam-hardening artifacts.

CT Scanning Protocol

Vendor-supplied software (CareDose4D; Siemens Healthcare, Erlangen, Germany) was used to determine patient size-specific tube current settings that would result in image quality comparable to that attained with a dose-reduced spiral unenhanced lung cancer screening protocol (tube voltage, 120 kVp; reference tube current–time product, 20 mAs for a typical 75-kg subject). Each patient underwent two dose-reduced spiral scans; an EID scan was followed by a PCD scan with identical tube voltage and tube current–time product after a 5–10-second delay to give breath-hold instructions. Spiral pitch was set at 0.8, with 0.5-second gantry rotation time for both systems. Because of the larger EID z-axis collimation (38.4 mm vs 16.0 mm), scanning duration was approximately 4.5 seconds for EID and 11.0 seconds for PCD. To examine image quality at different dose levels, a further dose-reduced single collimation axial (nonspiral) acquisition was performed with 100 kVp and 20 mAs (the lowest possible scanner setting) at the level of the carina (Fig E1 [online]). This scan was not adapted to individual body size, thus achieving a wider range of image noise and image quality. By using the same tube voltage and tube current–time product, volume CT dose index (CTDI_vol) estimates for the PCD system were 10%–12% higher than estimates for the EID system (Table 1). This is not a limitation of the PCD technology; it is due to the difference in z-axis collimation of the two detector systems, and this difference leads to different amounts of off-section scattered radiation, as explained by Dixon (14). Identical collimation would result in similar CTDI_vol values. We matched tube current–time product and tube voltage to ensure that both the number and the energy spectrum of x-ray photons incident on the two detector systems are similar to allow for a fair comparison between the two detector technologies. Effective dose was calculated by multiplying the dose-length product by 0.014 mSv/mGy·cm as the constant k value for chest imaging.

Table 1.

Subject Characteristics and Laboratory and Radiation Dose Parameters

Note.—Unless otherwise indicated, data are mean ± standard deviation (SD). Study subjects were divided into three groups based on body mass index (BMI): normal weight (BMI range, 18.5–24.9 kg/m²; group 1), overweight (BMI range, 25.0–29.9 kg/m²; group 2), and obese (BMI ≥30 kg/m², group 3). P values were calculated by using one-way analysis of variance.

*Data are numbers of patients, with percentages in parentheses.

^†Significant post hoc with Bonferroni correction versus group 1.

^‡Significant post hoc with Bonferroni correction versus group 2.

We scanned a 40-cm cylindrical phantom of water-equivalent material (QRM, Mohrendorf, Germany) with both detectors at multiple dose settings: 120 kVp and 20, 50, 100, and 200 mAs. Scans were performed twice at each setting, and the images were subtracted from each other to estimate de-trended image noise for noise power spectrum (NPS) calculations, as described by Friedman et al (15), by using a publicly available software package written for Matlab (Mathworks, Natick, Mass). In short, radial NPS was measured in 40 overlapping 128 × 128 pixel regions of interest (ROIs) distributed radially, with centers 7 cm off the isocenter of the detrended image.

CT Image Reconstruction

We used sinogram-affirmed iterative reconstruction (or SAFIRE) strength 2 with ReconCT (version 13.8.4.0; Siemens Healthcare). Sinogram-affirmed iterative reconstruction strength was based on American Association of Physicists in Medicine guidelines for lung cancer CT screening protocols (16). All data were reconstructed with 2-mm section thickness, 1-mm section increment, and a medium-smooth kernel (I31f) to assess soft tissue or a very sharp kernel (I70f) to assess the lungs and thoracic bones. Reconstructions were performed with a 27.5-cm field of view and 512 × 512 matrix.

Qualitative Image Analysis

Image quality was evaluated independently by three radiologists (E.C.J., L.R.F, A.A.M.; >20, >20, and 7 years of experience, respectively) using VuePACS software (version 12.0.0; Carestream Health, Rochester, NY). After a 4-week delay, one reader (E.C.J.) rescored all images to assess intrareader reproducibility. For a more detailed description of the qualitative image analysis with example images, see Appendix E1 (online).

Quantitative Image Analysis

Four circular ROIs of approximately 0.8–1.0 cm² were carefully positioned over the trachea, right and left lungs, and pectoral muscle (30 subjects × 4 ROIs × 2 dose settings = 240 pairs) (R.S., 5 years of experience). Image noise was measured as the SD of 1-cm² ROIs placed in a large uniform region of air in the trachea (17). For lung nodules, attenuation was measured by using a circular ROI of at least 10 mm², taking care to position ROIs in the same location for the EID and PCD images. ROI size was adapted to lung nodule size, with a margin of at least 1 mm to the outer surface of the nodule to limit partial volume effects. Lung nodule CNR was calculated as the absolute difference between the mean attenuation of the lung nodules and the surrounding normal-appearing lung tissue divided by image noise. Lung nodule size was measured by the three radiologists as the largest dimension visible on axial sections (18). The effects of beam-hardening and electronic noise artifacts were quantified by comparing the attenuation variation between an ROI placed within the paraspinal muscle at the level of the aortic arch, an area typically affected by beam hardening from the spine and scapulae, and an ROI placed within the pectoral muscle, an area unaffected by beam hardening. Attenuation stability was tested in the last nine patients by placing a calibrated test vial with an aqueous solution of an iodine-based contrast agent (iopamidol 300 mg/mL, Isovue 300; Bracco Diagnostics, Melville, NY) under the patient inside the field of view.

Statistical Analyses

Statistical analyses were performed by using R Statistical Software (version 3.3.1; Foundation for Statistical Computing, Vienna, Austria). Continuous data were expressed as mean ± SD. One-way analysis of variance was used to assess differences among BMI groups with Bonferroni-corrected Student t post hoc pairwise comparison. We used the paired t test to compare continuous variables and the paired Wilcoxon signed-rank test with continuity correction to compare qualitative image scores. The F test for equality of variance was used to compare the attenuation stability of both detectors. The McNemar test was used to compare diagnostic quality. Diagnostic quality was defined as a qualitative score of more than 1 on a four-point quality scale (1 = unacceptable, 2 = usable under limited conditions, 3 = probably acceptable, 4 = fully acceptable). Scan ranges are summarized in Figure E2 (online) (19). Inter- and intrareader reproducibility were assessed with the intraclass correlation coefficient (ICC). A two-way model with measures of consistency was used to calculate ICC values (20). Reproducibility was defined as poor (ICC <0.400), fair to good (ICC = 0.400–0.750), or excellent (ICC >0.750) (21). Two-sided P < .05 was considered to indicate a significant difference. Bonferroni-corrected P values were used to account for multiple ROI comparisons.

Sample size estimates were derived from the interstudy SD of image noise difference between EID and PCD dose-reduced scans at multiple different dose levels in a dedicated lung phantom by using the following equation, as described by Machin and Altman (9,22,23):

where α is the significance level, P is the study power, f is the value of the factor for different values of α and P, with σ as the interstudy SD, δ as the desired percentage difference to be detected, and n as the sample size needed. We deemed an image noise δ of 10% significant. To correct for increased variability in human experiments compared with previous phantom studies, we assumed σ was equal to 12.5%, which is twice the interstudy SD found in phantom experiments. These parameters resulted in a population of 30 subjects to detect 10% image noise difference with 85% power and α of .05.

Results

Baseline demographics of the study subjects and scanning parameters are summarized in Table 1. Ten subjects had a normal BMI (BMI range, 18.5–24.9 kg/m²), 10 were overweight (BMI range, 25.0–29.9 kg/m²), and 10 were obese (BMI ≥30.0 kg/m²). As expected, tube current–time product, CTDI_vol, dose-length product, and effective dose were significantly increased in the groups with higher BMI. CTDI_vol values were 10%–12% higher for PCD CT because of the different shape of the x-ray beam profile (see Materials and Methods).

Qualitative Image Analysis

Diagnostic quality of the PCD images was higher than that of the EID images (P = .016, P = .016, and P = .013 for readers 1, 2, and 3, respectively) (Table 2). Figure 1 shows quality scores for the dose-reduced scans assessed by the three readers, with example images in Figures 2–5. Image quality scores for assessment of lung tissue, lung nodules, soft tissue, and bone were higher for the PCD system, with lower subjective image noise when compared with those attained with the EID system (P < .001 for all). Beam-hardening artifacts were less pronounced on PCD images. The EID images showed fewer motion artifacts, likely due to the shorter breath-hold time (see Materials and Methods). Image quality scores for the different lung regions (apical, midlung, and base) are shown in Appendix E1 (online).

Table 2.

Diagnostic Quality of EID and PCD Dose-reduced CT

Note.—PCD scan quality was significantly better for each reader. P values were calculated with the McNemar test.

Figure 1:

Qualitative image scores for EID and PCD systems obtained by using 120-kVp dose-reduced scan settings. Scores were higher for lung tissue, lung nodules, soft tissue, bone image quality, image noise, and beam-hardening artifacts with the PCD system, whereas the EID system resulted in less streaking and fewer motion artifacts (all P < .001, paired Wilcoxon signed-rank test). Dark green indicates no artifact; green, mild artifact not interfering with diagnosis; light green, moderate artifact slightly interfering with diagnosis; yellow, pronounced artifact interfering with diagnosis; and red, artifact affecting the interpretation of a lesion or an organ of interest (based on European guidelines for image quality). For a more detailed description of the qualitative image analysis, with example images, see Appendix E1 (online).

Figure 2:

Example EID and PCD CT images in a 48-year-old man (window center, −500 HU; window width, 2000 HU). Axial, A, EID and, B, PCD reconstructed images at the level of the carina. Image noise was higher in A than in B (102.9 HU vs 83.2 HU); this is best seen in the posterior lung segments (arrowheads). Sagittal, C, EID and, D, PCD reconstructed images obtained through the right lung show there is less noise on D; this is best seen in the posterior lung regions (arrowheads).

Figure 5:

Example EID and PCD CT images in a 59-year-old man (window center, 490 HU; window width, 2500 HU). A, EID image shows low-attenuation areas in the paraspinal muscles (arrows) in the lung apices due to beam hardening. The cortex of a left upper rib (arrowheads) appears eroded. B, PCD image at the same level as A shows a more uniform appearance of the paraspinal muscles. The cortex of the upper left rib (arrowheads) is better visualized.

Figure 3:

Example EID and PCD CT images in a 68-year-old man (window center, −500 HU; window width, 2000 HU). Axial, A, EID and, B, PCD reconstructed images of the apical lung regions. Image noise was higher in A than in B (88.0 HU vs 73.1 HU); this is best seen in the posterior lung segments (arrowheads). Details of, C, axial EID and, D, PCD reconstructions highlight the greater PCD image quality.

Figure 4:

Example EID and PCD CT images in a 72-year-old man (window center, −500 HU; window width, 2000 HU). Axial, A, EID and, B, PCD reconstructed images at the level of the left lower lobe show a 5-mm incidental lung nodule (arrowhead). Details of, C, axial EID and, D, PCD reconstructions highlight greater lung nodule conspicuity and edge sharpness in D.

Intrareader agreement was high for all measures (all ICC ≥ 0.830) (Table 3). Similarly, interreader reproducibility was excellent for assessment of lung tissue image quality (ICC, 0.828), lung nodule image quality (ICC, 0.802), lung nodule edge sharpness (ICC, 0.892), lung nodule conspicuity (ICC, 0.873), soft-tissue image quality (ICC, 0.910), bone image quality (ICC, 0.831), qualitative image noise (ICC, 0.888), and image artifacts (ICC, 0.807).

Table 3.

Inter- and Intrareader ICC and 95% Confidence Interval

Note.—CI = confidence interval.

Quantitative Image Analysis

NPS curves for PCD CT were consistently lower than those for EID CT at all tube current levels and all spatial frequencies (Fig 6). The maximum difference was observed at the lowest setting (20 mAs), where the SD of noise, defined as the square root of the area under the NPS curve, was 12.3% lower for PCD. The SD of noise calculated from the NPS curves for PCD and EID, respectively, was 479.6 and 538.7 HU at 20 mAs, 289.1 and 323.3 HU at 50 mAs, 219.2 and 228.2 HU at 100 mAs, and 164.2 and 170.5 HU at 200 mAs.

Figure 6:

Graphs show the NPS for, A, EID and, B, PCD CT at 120 kVp and 20–200 mAs. PCD curves were consistently lower than EID curves. The difference is more prominent at lower tube currents, where electronic noise becomes more dominant in EID.

For 120-kVp reduced-dose spiral scans (Table 4), attenuation of the lungs, air, and pectoral muscle was similar between EID and PCD; differences were seen for air in the trachea (mean, −936.6 HU ± 17.2 for EID vs −942.2 HU ± 13.1 for PCD; P = .014) and paraspinal muscle (mean, 14.1 HU ± 25.3 for EID vs 41.4 HU ± 14.5 for PCD; P < .001). The attenuation difference between pectoral and paraspinal muscles was used as a metric to quantify beam-hardening artifacts (see Materials and Methods), and this difference was significantly larger for the EID (mean, 28.3 HU ± 26.8 vs 2.8 HU ± 12.4; P < .001). Mean attenuation variability for the test vial filled with a diluted iodine-based contrast agent was lower for PCD CT (130.5 HU ± 1.7) than for EID CT (120.0 HU ± 3.8) (P = .036).

Table 4.

Attenuation of ROIs at EID and PCD CT

Note.—Data are mean ± SD. For spiral acquisition, we performed a 120-kVp dose-reduced scan. For axial acquisition, we performed a 100-kVp dose-reduced scan.

For 100-kVp dose-reduced spiral scans (Table 4), attenuation values for EID images were more positive than those for PCD images in all ROIs; this was likely due to increased image noise.

PCD image noise was between 15.2% and 16.8% lower than EID image noise (77.4 HU ± 12.8 vs 89.2 HU ± 15.0 and 108.1 HU ± 21.8 vs 126.3 HU ± 25.5 for 120- and 100-kVp dose-reduced scans, respectively; P < .001 for both) in all BMI groups (Fig 7).

Figure 7:

Graph shows image noise values for EID and PCD systems at 120- and 100-kVp dose-reduced settings. Study subjects were divided into three groups based on BMI. PCD image noise was significantly lower than EID image noise for all BMI groups for both 120- and 100-kVp dose-reduced settings (P < .010 for all three BMI groups, paired t test).

Thirty-six incidental lung nodules (32 solid nodules, four nonsolid nodules) were available for comparison with the 120-kVp spiral scans. Lung nodule size (mean, 7.3 mm ± 4.5 for EID vs 7.1 mm ± 4.7 for PCD; P = .416) and attenuation (mean, 8.8 HU ± 31.0 for EID vs 6.7 HU ± 31.8 for PCD; P = .06) were similar for both detectors. Lung nodule CNR was 21.0% higher for PCD images than for EID images (12.1 ± 1.7 vs 10.0 ± 1.8, P < .001).

Discussion

To our knowledge, the value of PCD CT has not been previously evaluated for in vivo dose-reduced chest imaging. Previous phantom studies revealed lower image noise and better low-contrast resolution for PCD at multiple dose-reduced settings (10). Our results with dose-reduced PCD CT enabled in vivo confirmation of these findings in human subjects. Experienced readers identified the PCD images as having better diagnostic quality, with better lung nodule image quality, less image noise, and less beam hardening. Quantitative measurements showed that PCD images had 15.2%–16.8% lower noise at two different dose levels, with 21.0% higher lung nodule CNR. The Rose model for detection of low-contrast objects, such as lung nodules, states that object detection is based on object size, NPS, and CNR. Thus, we can anticipate that the increased lung nodule CNR and better noise power spectrum with photon-counting CT may improve lung nodule detection (24,25).

PCD CT has the unique characteristic of measuring both the energy and the number of photons (4). The better weighting of low-energy photons and the elimination of electronic noise may explain the improved attenuation stability and the reduced number of beam-hardening artifacts observed on PCD images. The improved attenuation stability leads to more reliable quantitative biomarkers in dose-reduced chest CT. This can be used to assess the evolution of diseases, such as emphysema, idiopathic pulmonary fibrosis, and α1-antitrypsin deficiency, in which attenuation changes of 1–3 HU may reflect disease progression (26). The reduced beam-hardening artifacts may improve image quality in regions prone to this artifact on conventional chest CT images, such as those obtained in the shoulder region (27).

Increasing concerns over radiation exposure from imaging tests and CT in particular have led to the rapid evolution of radiation dose reduction tools in the latest generations of CT scanners. Tools such as automated tube current modulation, tin filter, model-based iterative reconstruction, and corrections for clinical factors like thoracic diameter or breast size that have significantly reduced CT radiation doses while maintaining diagnostic image quality were not available for the prototype PCD system (28,29). A recent phantom study by Newell et al (30) using a third-generation dual-source scanner suggested that with the implementation of these tools, dose-reduced CT scans are possible at a CTDI_vol of 0.15 mGy. Because all these radiation dose reduction tools can be implemented on future iterations of photon-counting scanners, our results suggest that photon-counting technology may be an additional synergistic approach for CT radiation dose reduction.

This study had limitations. First, our study compared EID and PCD scans at two dose levels and may not reflect the performance of the system at different dose levels. However, we confirmed the feasibility of the PCD system at multiple dose levels in a phantom and did not consider it ethical to perform CT at any more dose levels in vivo. Second, we compared the PCD with a second-generation dual-source EID CT scanner and did not compare our results with those obtained with other state-of-the-art scanners. An important advantage of the current study design is that the same source x-ray spectra, focal spot size, beam filters, reconstruction kernels, system geometry, and artifact correction algorithms were used for both detector systems, and interscan variability (eg, due to patient positioning and motion) was minimized. Thus, the effects of nondetector components on image quality were minimized. However, our results might not be easily comparable with those obtained with other scanners. Third, we did not use tube current modulation to ensure consistent tube current values for both systems. This resulted in a limited number of diagnostically unacceptable images in the apical regions of larger patients. The EID system had 64 detectors in the z-axis direction, while the prototype PCD system had 32 detectors. Thus, breath-hold times with the PCD system were twice as long as those with the EID system, and readers detected more motion artifacts on PCD images. Fourth, charge-sharing in the PCD system could contaminate the low-energy (30–40-keV) photons. Modeling and correcting this artifact (eg, through coincidence detection) is an active area of research and may lead to further improvements in image quality (31,32). Fifth, further research is warranted to determine whether the improved image quality and lung nodule CNR attained with PCD CT can be translated into better diagnostic accuracy at dose-reduced chest CT.

In conclusion, this feasibility study with initial human results demonstrated good diagnostic quality, noise power spectrum, and lung nodule CNR of dose-reduced PCD chest CT when compared with conventional PCD CT.

Advances in Knowledge

• ■ Photon-counting detector (PCD) dose-reduced chest CT showed higher diagnostic quality (P = .016, P = .016, and P = .013 for readers 1, 2, and 3, respectively), with better subjective image quality scores for lung, soft tissue, and bone and less subjective image noise and fewer beam-hardening artifacts when compared with energy-integrating detector (EID) CT (P < .001 for all).

• ■ In objective image noise analysis, noise power spectrum showed up to 12.3% (479.6 HU vs 538.7 HU) lower noise for PCD CT than for EID CT over a range of radiation doses (20–200 mAs, 120 kVp) (P < .010).

Implications for Patient Care

• ■ PCD dose-reduced chest CT yielded better subjective and objective image quality when compared with radiation dose–matched EID CT.

• ■ For dose-reduced chest CT, PCD CT could improve diagnostic image quality, reduce radiation dose, or both, while maintaining diagnostic imaging quality.