Infant and toddler’s preoperative chest imaging on photon-counting detector CT: a paired-sample comparison with energy-integrating detector CT on image quality and radiation dose Authors: Xiang-Yu Liu, Wei Zhou, Ruo-Yang Shi, Wei-Hui Xie, Wei-Dong Qiao, Tang-Xu Sun, Sheng-Yang Xu, Ji Wang, Lian-Ming Wu Published: 2026 ABSTRACT AbstractBackgroundPhoton-counting CT (PCCT) may improve dose efficiency compared with energy-integrating detector CT (EID-CT), but evidence remains limited in very young children undergoing non-contrast preoperative chest CT under real-world clinical protocols.ObjectiveTo provide a size-aware matched comparison of radiation dose and image quality between PCCT and EID-CT in infants and toddlers (≤ 36 months) undergoing non-contrast preoperative chest CT, and to examine whether the main findings remained directionally consistent across prespecified strata of age, BMI, and water-equivalent diameter (DW).MethodsIn this single-center matched cohort study, a prospective PCCT cohort and a retrospective EID-CT cohort were propensity score–matched 1:1 (49 pairs). Dose metrics (CTDIvol, DLP, SSDE) and objective and subjective image quality were evaluated, with subgroup analyses stratified by age (< 12 vs. ≥ 12 months), BMI tertiles, and DW tertiles.ResultsThe PCCT group showed significantly lower radiation dose metrics (CTDIvol, DLP, and SSDE) compared to the EID-CT group (P < 0.001). In terms of objective image quality, PCCT significantly reduced image noise in all lung lobes and consistently demonstrated higher signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) (all P < 0.001). Subjective image quality assessments showed no significant difference between PCCT and EID-CT (P > 0.05). Across prespecified age, BMI, and DW strata, PCCT remained associated with lower dose metrics and favorable objective image-quality trends, although these subgroup findings should be interpreted as exploratory.ConclusionIn this matched cohort of infants and toddlers undergoing non-contrast preoperative chest CT, PCCT was associated with substantially lower radiation dose and improved objective image quality compared with EID-CT, and no statistically significant difference was detected in subjective image quality scores between PCCT and EID-CT.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12880-026-02339-7.Keywords: Photon-counting detector CT, Pediatric, Low-dose, Chest imaging, Propensity score matching Background Pediatric computed tomography (CT) utilization has increased over the last two decades [1–6]. This is especially concerning for children, who are more sensitive to radiation-induced carcinogenesis and have many remaining years of life for cancer to develop [6–8]. Although the focus on CT imaging in children has been reducing radiation dose, maintaining diagnostic image quality is paramount as well [9–11], especially for children who need essential preoperative examination and postoperative follow-up. Photon-counting detector CT (PCCT) is the new generation [12, 13] which differs from conventional energy-integrating detector CT(EID-CT) in the detector material and energy acquisition method and critically improves the spatial resolution, increases the utilization rate of radiation dose and reduces the radiation dose in some specific examinations [14]. Since each photon’s energy is individually measured and recorded, the energy information could be obtained for every scanning without specific setting [14–16]. This enhanced dose efficiency allows PCCT to deliver diagnostic image quality equivalent to that of EID-CT, but at significantly lower radiation doses, which is particularly beneficial in pediatric imaging. Moreover, photons with energy lower than 20 keV would be excluded from image reconstruction [17, 18] which effectively eliminates electronic noise and improve the resolution of low density tissues and also the image quality in low-dose imaging scenarios where the photon flux is reduced and electronic noise comprises a larger portion of the detected signal. This capability is particularly valuable in pediatric CT, where imaging protocols frequently rely on lower radiation doses compared to adults [19]. Numerous studies have highlighted its clinical advantages, such as enhanced dose efficiency and increased spatial resolution [20–22]. Preliminary pediatric thoracic studies have suggested that PCCT may reduce radiation dose while improving image quality compared with EID-CT; however, the generalizability of these findings remains limited by modest sample sizes and heterogeneous approaches to patient-size and dose assessment [23, 24]. This limitation is particularly relevant in infants and toddlers, in whom rapid variation in body size can complicate interpretation of scanner output and patient dose. Accordingly, the scientific aim of this study was not simply to add subgroup analyses, but to provide a size-aware matched comparison of PCCT and EID-CT in a narrowly defined and clinically relevant population—infants and toddlers (≤ 36 months) undergoing non-contrast preoperative chest CT before planned liver transplantation. Using age, BMI, DW, and SSDE, we sought to evaluate whether the dose-efficiency advantage of PCCT remains robust across the very small and rapidly changing body-size range characteristic of this population. Materials and methods Study population and cohort assembly This study comprised two cohorts: a prospective cohort undergoing PCCT and a retrospective cohort undergoing EID-CT. The prospective cohort included consecutive pediatric patients with a clinical indication for preoperative assessment prior to planned liver transplantation, who were referred for chest CT on a PCCT scanner between July 2024 and January 2025. The retrospective cohort included consecutive pediatric patients meeting the same clinical indication, who underwent non-contrast chest CT using an EID-CT scanner between January 2024 and June 2024. Eligibility criteria for cohorts were: (1) age ≤ 36 months at the time of CT, (2) successful completion of the examination, and (3) provision of written informed consent for the prospective cohort. Exclusion criteria were: (1) use of intravenous contrast medium, (2) multi-regional chest CT protocols, and (3) incomplete baseline demographic or anthropometric data. The patient selection, exclusion process, and final matched-cohort assembly are summarized in Fig. 1. Study flow diagram for cohort selection and propensity score matching. PCCT, photon-counting detector computed tomography; EID-CT, energy-integrating detector computed tomography; BMI, body mass index; DW, water-equivalent diameter Initially, 109 patients were identified in each cohort. For each patient, baseline variables included sex, age, height, weight, body mass index (BMI), water-equivalent diameter (DW), and clinical indication. Age, BMI, and DW were prespecified for propensity score matching and stratified analyses, whereas all baseline variables were used to describe the matched cohorts. The study protocols for both cohorts were approved by the Institutional Review Board (IRB, No.LY2024-036-B). For the prospective cohort, informed consent was obtained from the legal guardians of any participant. For the retrospective cohort, the requirement for written informed consent was waived by the IRB. CT acquisition protocols Both PCCT and EID-CT examinations were performed using vendor-recommended protocols, with protocol settings optimized in accordance with the ALARA principle. All scans were acquired with patients in the supine position, arms raised above the head, and in a craniocaudal direction from the lung apices to the lung bases. No sedation was administered prior to scanning; all examinations were performed during quiet breathing. PCCT examinations were performed on a NAEOTOM Alpha scanner (Siemens Healthineers, Forchheim, Germany) using automated tube current modulation (CARE Dose4D, Siemens Healthineers) to determine the tube current–time product (mAs) with an image quality (IQ) level of 80 at 90 kV in Quantum mode. Collimation was 144 × 0.4 mm, with a rotation time of 0.25 s and a pitch of 1.5. EID-CT examinations were performed on a Dual-Layer Spectral CT 7500 scanner (Philips Medical Systems, Netherlands) using vendor-provided Quality Reference mAs (DoseRight, Philips Medical Systems) with an index of 10 at 80 kV. Collimation was 128 × 0.625 mm, with a rotation time of 0.27 s and a pitch of 1.5. At our institution, sharp lung kernels are routinely used for preoperative pediatric chest CT to enhance visualization of lung parenchymal detail. Axial images from both systems were reconstructed with a 1.0-mm slice thickness using sharp lung convolution kernels (Bl60 for PCCT and Y-Detail for EID-CT). PCCT datasets were reconstructed with quantum iterative reconstruction (QIR; Siemens Healthineers) at strength level 3 (of 4), with virtual monoenergetic images generated at 70 keV. EID-CT datasets were reconstructed with iDose (Philips Medical Systems) at strength level 3 (of 4). These reconstruction settings were selected by consensus of pediatric thoracic radiologists as a pragmatic balance between noise reduction and preservation of image texture. Radiation dose metrics were extracted from the dose reports archived in the picture archiving and communication system (PACS) and included the volume CT dose index (CTDIvol), dose–length product (DLP), size-specific dose estimate (SSDE) and water-equivalent diameter (DW). Because the two examinations were performed using clinically implemented vendor-recommended protocols on different scanner platforms, the present comparison should be interpreted as a pragmatic protocol-level comparison rather than as an isolated evaluation of detector technology alone. Objective image quality Quantitative image analysis was independently conducted by two radiologists (W.Z. and W.D.Q) with 22 and 7 years of experience in chest CT, respectively. Both readers were blinded to the CT system used and performed measurements using semi-automated software (Syngo.via, version VB40; Siemens Healthineers). PCCT and EID-CT examinations were anonymized and randomly assigned to the readers. Circular regions of interest (ROIs) were manually placed on 1.0-mm-thick axial images in normal lung parenchyma within the right upper lobe (RUL), right middle lobe (RML), right lower lobe (RLL), left upper lobe (LUL), and left lower lobe (LLL), avoiding bronchi, vessels, and other large anatomic structures. For contrast-to-noise ratio (CNR) measurements, an additional ROI was placed in the subcutaneous fat at the level of the fifth thoracic vertebra (T5) as background reference values. All ROIs had a standardized area of 1 cm². When this was not feasible without including abnormal parenchyma or large structures, the largest possible ROI was used. For each ROI, mean attenuation and standard deviation (SD, both in Hounsfield units) were recorded, representing lung attenuation and image noise, respectively. Three ROIs were placed in each lobe, and their mean attenuation and noise values were calculated. The signal-to-noise ratio (SNR) and CNR were calculated with equations below: Subjective image quality Subjective image quality was independently evaluated by two additional radiologists (W.H.X and R.Y.S), with 8 and 11 years of experience in chest CT, respectively. The readers were blinded to patient demographic information and the CT system used. All CT images were reviewed on a Picture Archiving and Communication System (PACS) workstation (Vue, version 11.4, Carestream Health; Philips Medical Systems). For image evaluation, the lung and mediastinal window settings were initially adjusted as follows: window width of 1400 HU and window level of − 400 HU for the lung window, and window width of 360 HU and window level of 40 HU for the mediastinal window. Readers were permitted to further adjust window settings at their discretion. Each image set was assessed for two subjective image quality parameters: overall image quality and image noise. Overall image quality was defined as the general diagnostic acceptability of the image based on the clarity of anatomical structures, absence of artifacts, and adequacy for clinical interpretation, and was graded using a 5-point Likert scale: 1, poor, nondiagnostic; 2, fair; 3, moderate; 4, good; and 5, excellent. Image noise was defined as the random variation of pixel values that obscures fine anatomical details or reduces diagnostic clarity, and was graded using a 5-point Likert scale: 1, severe, nondiagnostic; 2, substantial; 3, moderate; 4, minimal; and 5, none. Statistical analysis Propensity score matching (PSM) was performed using R software (version 4.4.1; R Foundation for Statistical Computing) with a 1:1 nearest-neighbor algorithm, matching on patient age in month, BMI, and DW. Continuous variables were assessed for normality using the Kolmogorov–Smirnov test and reported as mean ± standard deviation (SD) for normally distributed data or median with interquartile range (IQR) for non-normally distributed data. Categorical variables were summarized as counts and percentages, and group differences were assessed using Fisher’s exact test, as appropriate. The distribution of differences between matched pairs was evaluated with the Shapiro–Wilk test. Depending on normality, paired Student’s t-tests or Wilcoxon signed-rank tests were used to compare quantitative parameters and radiation dose metrics between groups. Subgroup analyses stratified by age, BMI, and DW were performed using the same statistical approach, with stratification as follows: Age: <12 months and ≥ 12 months; BMI: tertiles based on the sample distribution; DW: tertiles based on the sample distribution. Inter-rater agreement for continuous quantitative measurements (e.g., CT values, CNR, and SNR) was assessed using the intraclass correlation coefficient (ICC) with a two-way random-effects model for absolute agreement [ICC(2,1)]. The ICC values were interpreted as follows: <0.40, poor; 0.40–0.59, fair; 0.60–0.74, good; and ≥ 0.75, excellent agreement. Likert scores for the subjective image quality measures were summarized for each reader as median with IQR and were compared between groups using the Wilcoxon signed rank test. Interobserver agreement between the two readers for the subjective image quality measures was calculated using weighted kappa coefficients. A two-sided P value of < 0.05 was considered statistically significant. To address multiplicity, the primary matched-cohort comparisons were prespecified as the three radiation dose metrics and the fifteen lobe-based objective image-quality metrics. P values for these primary comparisons were adjusted using the Holm method. Subjective image-quality analyses were interpreted separately. Stratified analyses by age, BMI, and DW were considered exploratory and were interpreted cautiously rather than as confirmatory tests. Results Cohort matching and baseline balance After application of the eligibility and exclusion criteria, 108 patients remained in the PCCT cohort and 106 in the EID-CT cohort. After 1:1 propensity score matching, 49 patients remained in each group. Figure 2 shows the standardized mean differences before and after matching, confirming improved balance for the prespecified matching variables. After matching, the standardized mean differences for age, BMI, and DW were 0.011, 0.136, and 0.089, respectively. SMDs before and after PSM. PSM, Propensity score matching; SMDs, standardized mean differences Table 1 summarizes the baseline characteristics of the matched cohorts. The PCCT group included 26 girls and 23 boys, whereas the EID-CT group included 27 girls and 22 boys. No significant between-group differences were observed in median age (6 vs. 6 months, P = 0.783), BMI (15.5 vs. 15.7, P = 0.563), height (67.5 vs. 66.5 cm, P = 0.649), weight (7.2 vs. 7.1 kg, P = 0.765), or DW (13.6 vs. 13.7, P = 0.624). Table 1Patient characteristicsVariablesTotal (n = 98)EIDCT (n = 49)PCCT (n = 49) P Age (Months)6.00 (5.00, 9.00)6.00 (5.00, 9.00)6.00 (5.00, 8.00)0.783Gender (%)1.00Male45 (45.91%)22 (44.90%)23 (46.94%)Female53 (54.08%)27 (55.10%)26 (53.06%)BMI (kg/m²)15.65 (14.65, 16.75)15.62 (14.65, 17.01)15.69 (14.65, 16.57)0.563Height (cm)65.00 (62.00, 68.75)65.00 (61.00, 68.00)65.00 (62.00, 69.00)0.649Weight (kg)6.57 (5.62, 8.20)6.50 (5.70, 8.10)6.70 (5.60, 8.30)0.765DW (cm)13.70 (12.83, 14.30)13.70 (12.90, 14.30)13.70 (12.70, 14.30)0.624Clinical diagnosis (%)0.449aBiliary atresia87 (88.78%)45 (91.84%)42 (85.71%)Cholestasis3 (3.06%)2 (2.04%)1 (1.02%)Hepatoblastoma3 (3.06%)1 (1.02%)2 (2.04%)Langerhans cell histiocytosis1 (1.02%)01 (1.02%)Budd–Chiari syndrome2 (2.04%)02 (2.04%)Hepatolenticular degeneration1 (1.02%)01 (1.02%)Alagille syndrome1 (1.02%)1 (1.02%)0Data are presented as median (25th quantile, 75th quantile) or n(%); aFisher’s exact tests Patient characteristics Data are presented as median (25th quantile, 75th quantile) or n(%); aFisher’s exact tests Clinical indications were also broadly comparable between groups and were dominated by biliary atresia. In the PCCT group, indications included biliary atresia (n = 42), cholestasis (n = 1), hepatoblastoma (n = 2), Langerhans cell histiocytosis (n = 1), Budd–Chiari syndrome (n = 2), and hepatolenticular degeneration (n = 1). In the EID-CT group, indications included biliary atresia (n = 45), cholestasis (n = 2), hepatoblastoma (n = 1), and Alagille syndrome (n = 1). Interobserver agreement Objective measurements including CT attenuation of the five pulmonary lobes(LUL, LLL, RUL, RML, RLL) and subcutaneous fat at the T5 level showed excellent reproducibility. The ICCs were 0.776, 0.812, 0.756, 0.899, 0.781 and 0.838, respectively (all P < 0.05), confirming high interobserver reliability across all parameters. Radiation doses Table 2; Fig. 3 show that all radiation dose metrics were significantly lower in the PCCT group than in the EID-CT group. All primary matched-cohort dose findings remained significant after Holm adjustment. Specifically, the mean CTDIvol was 0.49 ± 0.17 mGy for PCCT versus 2.09 ± 0.38 mGy for EID-CT, yielding a mean difference of − 1.60 mGy (95% CI, − 1.72 to − 1.48; t = − 27.45, df = 48, P < 0.001). The mean DLP was 9.41 ± 3.86 mGy·cm for PCCT versus 54.98 ± 9.06 mGy·cm for EID-CT, with a mean difference of − 45.57 mGy·cm (95% CI, − 48.03 to − 43.10; t = − 37.20, df = 48, P < 0.001). Similarly, the mean SSDE was 1.18 ± 0.37 mGy for PCCT versus 4.69 ± 0.86 mGy for EID-CT, corresponding to a mean difference of − 3.51 mGy (95% CI, − 3.78 to − 3.25; t = − 26.39, df = 48, P < 0.001). Table 2Comparison of radiation dose between two groupsParameterPCCT (Mean ± SD)EID-CT (Mean ± SD) P CTDIvol (mGy)0.49 ± 0.172.09 ± 0.38< 0.001DLP (mGy·cm)9.41 ± 3.8654.98 ± 9.06< 0.001SSDE (mGy)1.18 ± 0.374.69 ± 0.86< 0.001Data are presented as mean ± SD; PCCT, photon-counting CT; EID-CT, energy-integrating detector CT. CTDIvol= volume computed tomography dose index; DLP= dose–length product; SSDE=size-specific dose estimate Comparison of radiation dose between two groups Data are presented as mean ± SD; PCCT, photon-counting CT; EID-CT, energy-integrating detector CT. CTDIvol= volume computed tomography dose index; DLP= dose–length product; SSDE=size-specific dose estimate Paired comparison of radiation dose metrics between PCCT and EID-CT in the matched cohort. PCCT, photon-counting detector computed tomography; EID-CT, energy-integrating detector computed tomography; CTDIvol, volume computed tomography dose index; DLP, dose–length product; SSDE, size-specific dose estimate. ***P < 0.001 Objective image quality Table 3; Fig. 4 show that PCCT consistently yielded lower image noise and higher SNR/CNR than EID-CT across all five pulmonary lobes. All primary matched-cohort objective image-quality findings remained significant after Holm adjustment. Mean SD on PCCT ranged from 48.88 to 51.10 HU, whereas EID-CT ranged from 62.86 to 72.57 HU, yielding absolute reductions of 12.80–21.89 HU (approximately 20–30% lower noise, depending on lobe). The largest noise reduction was observed in the left upper lobe (LUL: 50.69 ± 9.36 HU vs. 72.57 ± 12.48 HU; mean difference − 21.89 HU; P < 0.001). PCCT also demonstrated higher SNR and CNR in every lobe (all P < 0.001). Mean SNR increases ranged from + 2.92 to + 4.74 (approximately + 26% to + 48% relative increase across lobes). Mean CNR increases ranged from + 4.04 to + 4.76 (approximately + 44% to + 51% relative increase). For example, in the right upper lobe, SNR was 15.13 ± 4.73 on PCCT versus 10.66 ± 2.43 on EID-CT (mean difference + 4.74; P < 0.001), and CNR was 13.97 ± 4.03 versus 9.62 ± 2.54 (mean difference + 4.35; P < 0.001).(Fig. 4). Table 3Comparison of objective image quality between two groupsParameterPCCT (Mean ± SD)EID-CT (Mean ± SD) P LUL-SD50.69 ± 9.3672.57 ± 12.48< 0.001LUL-SNR14.29 ± 3.689.64 ± 2.23< 0.001LUL-CNR14.20 ± 4.169.43 ± 2.44< 0.001LLL-SD50.69 ± 9.3565.22 ± 11.45< 0.001LLL-SNR13.44 ± 3.5610.47 ± 2.14< 0.001LLL-CNR13.29 ± 4.089.25 ± 2.58< 0.001RUL-SD48.88 ± 11.4566.47 ± 11.50< 0.001RUL-SNR15.13 ± 4.7310.66 ± 2.43< 0.001RUL-CNR13.97 ± 4.039.62 ± 2.54< 0.001RML-SD50.06 ± 9.1962.86 ± 11.70< 0.001RML-SNR14.22 ± 3.5211.31 ± 2.17< 0.001RML-CNR13.89 ± 4.339.61 ± 2.07< 0.001RLL-SD51.10 ± 7.8167.35 ± 14.59< 0.001RLL-SNR13.17 ± 2.9710.20 ± 3.08< 0.001RLL-CNR13.23 ± 4.428.91 ± 1.99< 0.001Data are presented as mean ± SD; PCCT, photon-counting CT; EID-CT, energy-integrating detector CT. RUL, right upper lobe, RML, right middle lobe, RLL, right lower lobe, LUL, left upper lobe, LLL, left lower lobe, CNR, contrast-to-noise ratio, SD, standard deviation, SNR, signal-to-noise ratio Comparison of objective image quality between two groups Data are presented as mean ± SD; PCCT, photon-counting CT; EID-CT, energy-integrating detector CT. RUL, right upper lobe, RML, right middle lobe, RLL, right lower lobe, LUL, left upper lobe, LLL, left lower lobe, CNR, contrast-to-noise ratio, SD, standard deviation, SNR, signal-to-noise ratio Box plots depict the comparative distribution of objective image quality metrics between PCCT and EID-CT. PCCT, photon-counting CT; EID-CT, energy-integrating detector CT. LUL, left upper lobe; LLL, left lower lobe; RUL, right upper lobe; RML, right middle lobe; RLL, right lower lobe; SD, standard deviation; CNR, contrast-to-noise ratio; SNR, signal-to-noise ratio. *P < 0.05; **P < 0.01; ***P < 0.001 Subjective image quality Overall percent agreement between the two readers was 76% (37/49) for overall image quality in the PCCT group and 73% (36/49) in the EID-CT group. For subjective noise, overall percent agreement was 82% (40/49) and 88% (43/49), respectively. Interreader agreement was both excellent for overall image quality (weighted κ = 0.73) and noise (weighted κ = 0.81). Table 4 summarizes the subjective scores for both readers. There was no significant difference in median overall image quality between PCCT and EID-CT for reader 1 (5 vs. 5, P = 0.65) or reader 2 (5 vs. 5, P = 0.47). Similarly, median noise scores did not differ significantly between groups for either reader (all P > 0.05). Representative cases spanning a range of subjective scores are provided in Fig. 5. Table 4Comparison of subjective image quality measures for PCCT and EID-CTMeasure, reader, scorePCCTEID-CT P a Overall quality Reader 10.65Excellent (score = 5)3735Good (score = 4)78Moderate (score = 3)56Fair (score = 2)00Poor (score = 1)00Median score (IQR)5 (5–5)5 (4–5)Reader 20.47Excellent (score = 5)4037Good (score = 4)68Moderate (score = 3)34Fair (score = 2)00Poor (score = 1)00Median score (IQR)5 (5–5)5 (5–5) Image noise Reader 10.78None (score = 5)2523Minimal (score = 4)2023Moderate (score = 3)31Substantial (score = 2)12Severe (score = 1)00Median score (IQR)5 (4–5)4 (4–5)Reader 20.86None (score = 5)2526Minimal (score = 4)2120Moderate (score = 3)12Substantial (score = 2)21Severe (score = 1)00Median score (IQR)5 (4–5)5 (4–5)Except where otherwise indicated, data are number of CT examinations. PCCT = photon-counting detector, EID-CT = energy-integrating detector. aP values from paired Wilcoxon signed-rank tests Comparison of subjective image quality measures for PCCT and EID-CT Except where otherwise indicated, data are number of CT examinations. PCCT = photon-counting detector, EID-CT = energy-integrating detector. aP values from paired Wilcoxon signed-rank tests Different location of axial CT images of chest in two matched patients; both examinations were performed owing to preoperative chest CT scan. Photon-counting detector CT (PCCT) image of 6-month-old patient. For PCCT examination, CTDIvol was 0.36 mGy, DLP was 8.4 mGy·cm, and SSDE was 0.71 mGy. Energy-integrating detector CT (EID-CT) image of 7-month-old patient. For EID-CT examination, CTDIvol was 1.87 mGy, DLP was 46.8 mGy·cm, and SSDE was 3.95 mGy. For overall image quality, both readers assigned score of 4 (good) for PCCT and score of 4 (good) for EID-CT. For noise, both readers assigned score of 5 (minimal) for PCCT and score of 4 (none) for EID-CT. Representative CT number and SD from these ROIs are shown for illustration Stratified exploratory analyses Supplemental Tables 1–3 and Supplemental Figs. 1–3 summarize the exploratory analyses stratified by age, BMI, and DW. Subgroup comparisons are reported as paired mean differences with corresponding 95% CIs. Overall, the favorable trends observed in the primary matched-cohort analysis remained directionally consistent across strata, with lower radiation dose and generally improved objective image quality in the PCCT group. However, some subgroup-specific lobe-level comparisons were imprecise and did not reach statistical significance, particularly in the smaller strata, and these findings should therefore be interpreted as exploratory rather than confirmatory. Discussion Principal findings This study compared non-contrast preoperative chest CT in infants and toddlers scanned on PCCT versus EID-CT system using a matched-cohort design. We observed markedly lower radiation dose surrogates (CTDIvol, DLP, and SSDE) with PCCT together with improved objective lung image quality (lower noise and higher SNR/CNR). In contrast, no statistically significant difference was detected in reader-based subjective image quality scores between the two systems. Notably, the largest relative SSDE reductions were observed in the smallest-patient strata (BMI < 14.6 kg/m2 or DW < 12.8 cm), where SSDE was reduced by more than 80% compared with EID-CT. These findings should be interpreted within the specific scanner-, protocol-, reconstruction-, and energy-domain pairing used at our institution, and should therefore be viewed as a pragmatic real-world comparison rather than a detector-isolated experiment. Comparison with previous studies Our results align with and extend prior pediatric PCCT thoracic studies. Initial clinical experience has demonstrated the feasibility of ultra-low-dose pediatric chest imaging on PCCT with good-to-excellent visualization of lung structures [23], and matched pediatric comparisons have reported significant dose reductions with PCCT while maintaining overall diagnostic acceptability for high-resolution chest CT [24]. In addition, a paired-scanner study in children undergoing contrast-enhanced chest CT found lower radiation exposure and improved objective lung image quality with PCCT, while overall subjective image quality did not significantly differ [25]. Compared with these studies, our cohort was younger and focused on a preoperative, non-contrast clinical indication, which were more sensitive to radiation dose than those in previous studies. Our stratified analyses should be interpreted as exploratory assessments of robustness rather than as stand-alone evidence of effect modification. They provide additional clinical context by showing that the favorable dose and objective image-quality trends observed in the matched cohort remained directionally consistent across age-, BMI-, and DW-defined strata, with the largest relative SSDE reductions observed in the smallest body-size categories. However, because these subgroup analyses involved substantially more comparisons and several strata contained relatively small numbers of matched pairs, subgroup-specific nominal P values should be interpreted cautiously. Interpretation of body-size–stratified findings The lower dose metrics observed with PCCT likely reflect a combination of detector physics and protocol implementation, but these contributions cannot be separated in a two-scanner comparison. PCCT can improve dose efficiency by reducing electronic noise contribution and enabling energy-resolved signal formation. Technical reviews emphasize that these properties can translate into improved contrast-to-noise performance and dose reduction compared with conventional energy-integrating detectors, especially in low-dose settings [26, 27]. In our cohort, these potential advantages were reflected in lower SSDE at matched clinical image quality, suggesting that PCCT may provide an opportunity to reduce dose for preoperative pediatric chest CT without sacrificing interpretability. However, the magnitude of benefit is expected to depend strongly on the chosen acquisition mode, tube potential, automatic exposure control settings, and reconstruction strategy. Technical considerations and sources of image-quality advantage Objective lung image quality improvements in our cohort likely also depended on reconstruction choices. We reconstructed both systems with sharp lung kernels and used QIR at strength level 3 on PCCT. Prior work has shown that QIR reduces noise and improves CNR, and that the magnitude of improvement can be kernel dependent with sharper kernels benefiting more. In lung-focused studies, QIR level 3 has been reported as a favorable compromise between noise suppression and preservation of fine structural detail [28]. Similar observations have been reported in other PCCT applications, where QIR improves noise and CNR and average subjective preference often peaks around QIR level 3, particularly for sharper kernels [29]. Accordingly, the objective advantage observed in our PCCT reconstructions should be interpreted as the combined effect of detector performance and reconstruction, rather than a detector-only phenomenon. Methodological limitations of the cross-platform comparison A key methodological consideration is the image (energy) domain used for comparison. In our study, objective metrics on PCCT were derived from 70-keV virtual monoenergetic images (VMIs), consistent with commonly used clinical PCCT workflows for non-contrast examinations [30]. By contrast, the EID-CT reference images were polychromatic 80-kVp reconstructions. Although spectral detector–based CT can generate monoenergetic images, the dual-layer Philips platform requires acquisition at 120 or 140 kVp to obtain spectral data, whereas lower kVp settings (e.g., 80 kVp) are used to generate conventional images only. In infants and toddlers, pediatric dose optimization frequently relies on low-kVp techniques, and increasing tube voltage increases radiation output at matched tube current; therefore, performing an additional 120-kVp spectral acquisition solely to enable VMI matching would conflict with ALARA and was not ethically appropriate in our cohort. Accordingly, our comparison should be interpreted as a pragmatic, real-world evaluation of vendor-recommended protocols under clinical constraints rather than an idealized, energy-domain–matched experiment. Finally, energy selection and monoenergetic reconstruction can alter attenuation, noise magnitude, and noise texture, which may partially contribute to differences in noise, SNR, and CNR between systems [19]. Future work should evaluate whether similar conclusions hold when using low-threshold reconstructions (e.g., T3D) or when matching reconstruction domains more closely (polychromatic-equivalent images versus polychromatic, or keV-matched VMIs when feasible). Clinical implications and future directions The absence of a difference in subjective image quality despite improved objective metrics is clinically informative. It suggests that our institutional EID-CT protocol already provides task-adequate image quality for preoperative chest CT, and that the objective advantage observed with PCCT may represent “headroom” that could be exchanged for lower dose, greater robustness to motion, or—under selected conditions—shorter acquisition times without compromising reader perception. High-pitch acquisition has been reported to reduce motion artifacts and, in certain pediatric free-breathing settings, may lower radiation exposure [31]. A pediatric phantom study further demonstrated that PCCT can support very-low-dose high-pitch lung imaging with adequate image quality, particularly when combined with higher QIR strengths [32]. However, extrapolating these strategies to our cohort of very young infants and toddlers warrants caution. Given the very short scan length in this age group, the absolute time savings of high-pitch modes may be modest, whereas the substantially higher table speed may generate mechanical vibration and audible cues that can arouse lightly sleeping, non-sedated infants in real-world practice, potentially triggering abrupt motion and increasing the risk of repeat scanning. Moreover, overranging and related effects may increase DLP at very high pitch [33]. Accordingly, we believe that the application of UHR and high-pitch strategies in extremely small body sizes remains an important topic for future prospective studies, ideally incorporating task-based diagnostic endpoints, motion-robustness metrics, and workflow-relevant outcomes (including the feasibility of non-sedated imaging). Limitation This study has several limitations. It was performed at a single institution with a modest sample size, and the matched cohorts were acquired on different scanners rather than via intra-individual repeat imaging; residual confounding related to patient condition, respiratory phase, or selection bias may remain. Protocol harmonization was constrained by system differences (tube potential options and reconstruction domains), and proprietary reconstruction algorithms limited isolation of detector versus reconstruction contributions. We focused on quantitative and reader-based image quality metrics rather than task-based diagnostic performance, and we did not assess lesion detectability or clinical outcomes. Finally, the study evaluated only non-contrast preoperative chest CT acquired in Quantum mode with one reconstruction configuration; future multicenter studies should investigate additional acquisition modes, reconstruction domains, and task-based endpoints to establish generalizable pediatric PCCT protocol recommendations across indications and patient sizes. Conclusion In this single-center propensity score–matched cohort of infants and toddlers (≤ 36 months) undergoing non-contrast preoperative chest CT, PCCT was associated with markedly lower radiation dose and better objective image quality than EID-CT, while no statistically significant difference was detected in reader-based subjective image quality scores between the two systems. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgments We thank the pediatric radiology team and CT technologists at Renji Hospital for their support in patient recruitment and image acquisition. We are also grateful to the participating children and their families for making this study possible. Abbreviations As low as reasonably achievable Body mass index Contrast-to-noise ratio Computed tomography Computed tomography dose index Volume computed tomography dose index Dose–length product Water-equivalent diameter (D_W) Energy-integrating detector computed tomography Hounsfield unit Intraclass correlation coefficient Interquartile range Left lower lobe Left upper lobe Milligray Photon-counting detector computed tomography Propensity score matching Quantum iterative reconstruction Right lower lobe Right middle lobe Right upper lobe Signal-to-noise ratio Size-specific dose estimate Ultra-high resolution Virtual monoenergetic image Author contributions [Xiang-Yu Liu]: Conceptualization, Methodology, Investigation, Writing – original draft. [Wei Zhou]: Data curation, Formal analysis, Visualization. [Ruo-Yang Shi]: Formal analysis [Wei-Hui Xie]: Formal analysis [Wei-Dong Qiao]: Investigation [Tang-Xu Sun]: Investigation [Sheng-Yang Xu]: Data curation [Ji Wang]: Resources, Supervision [Lian-Ming Wu]: Funding acquisition, Supervision, Writing – Review & Editing. Xiang-Yu Liu and Wei Zhou contributed equally as co-first authors. Ji Wang and Lian-Ming Wu jointly supervised as co-corresponding authors. All authors read and approved the final manuscript. Funding This study has received funding by National Youth Talent Support Program; National Natural Science Foundation of China (82171884, 82471931); Shanghai Municipal Commission of Science and Technology Medical Innovation Research Special Project (23Y11906900); Shanghai “Yiyuan New Star” Outstanding Youth Talent (Excellent Program). Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Renji Hospital (LY2024-036-B). This observational matched-cohort imaging study was not prospectively registered. Written informed consent was obtained from legal guardians of all subjects (patients) of in this study. Generative AI use The authors of this manuscript declare no generative AI use in this manuscript. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Xiang-Yu Liu and Wei Zhou contributed equally to this work. Ji Wang and Lian-Ming Wu contributed equally as corresponding authors. Contributor Information Ji Wang, Email: jeff1885@126.com. Lian-Ming Wu, Email: wlmssmu@126.com. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.