Size-specific dose estimation in pediatric computed tomography: From dose characterization to individualized optimization

Abstract

Pediatric computed tomography (CT) is clinically indispensable for emergency care, neurologic disorders, thoracoabdominal diseases, trauma assessment, and oncologic diagnosis and follow-up. However, children should not be regarded as small adults. They are actively growing and developing, and organs or tissues such as the hematopoietic system, thyroid, breast, gonads, lens, central nervous system, and bone marrow are more sensitive to ionizing radiation. In addition, children have a longer life expectancy, allowing a longer latency window for radiation-related late effects. Epidemiological studies suggest associations between pediatric CT exposure and cumulative-dose-related risks of leukemia, brain tumors, and overall cancer; although the individual absolute risk is generally low, stricter justification, optimization, and dose recording remain necessary in pediatric populations. Conventional volume CT dose index (CTDI_vol) and dose-length product primarily describe scanner output under standard phantom conditions and cannot adequately characterize differences in patient body size, tissue attenuation, organ location, and scan coverage. Size-specific dose estimate (SSDE), which corrects CTDI_vol using a patient-size conversion factor, represents an important intermediate dose descriptor linking scanner output, pediatric body-size characteristics, organ-dose estimation, and scan-protocol optimization. This review focuses on the specific requirements of pediatric CT dose assessment. It summarizes the conceptual evolution of SSDE, the selection of size metrics, applications across examination sites, relationships with organ dose and radiation-risk assessment, pediatric diagnostic reference levels, and future directions in automated dose management. The aim is to provide a conceptual and practical basis for individualized pediatric CT dose optimization.

Key Words: Children; Computed tomography; Size-specific dose estimate; Water-equivalent diameter; Organ dose; Diagnostic reference level; Dose optimization

Core Tip: Pediatric computed tomography (CT) dose assessment should move beyond scanner-output indices toward patient-size-adapted optimization. This review highlights size-specific dose estimate (SSDE) as a practical bridging metric linking volume CT dose index correction, pediatric body-size variation, water-equivalent diameter, head-specific dose estimation, organ-dose approximation, diagnostic reference levels, and task-based image-quality optimization. By integrating SSDE with automated size extraction, organ-dose modeling, and clinical indication-specific protocols, pediatric CT dose management can shift from simple dose recording to individualized, quality-constrained optimization.

INTRODUCTION

Computed tomography (CT) offers rapid acquisition, high spatial resolution, and broad applicability in emergency settings, making it an important imaging technique for pediatric trauma, neurological emergencies, pulmonary infection, abdominal pain, tumor follow-up, and evaluation of complex congenital diseases. However, CT uses ionizing radiation, and radiation dose management in children requires greater attention than in adults. Pediatric tissues have active cellular proliferation, and some organs remain developmentally immature; therefore, the potential biological consequences of radiation-induced DNA damage and long-term effects warrant particular attention. Children also have a longer life expectancy, allowing radiation-associated malignancies more time to manifest. The retrospective cohort study by Pearce et al[1] in the United Kingdom showed that the risks of leukemia and brain tumors after CT exposure in children and young adults were associated with absorbed dose; Mathews et al[2], using data linkage from 11 million Australians, further suggested an increased overall cancer incidence after CT exposure during childhood or adolescence; and Miglioretti et al[3] estimated organ doses and attributable cancer risks associated with pediatric CT use across multiple healthcare systems in the United States. These studies do not imply that clinically necessary CT examinations should be avoided; rather, they emphasize that pediatric CT must adhere to the principles of justification and individualized dose optimization.

Conventional CT dose metrics include volume CT dose index (CTDI_vol) and dose-length product (DLP). Both are useful for monitoring scanner output, comparing protocols, and establishing diagnostic reference levels (DRLs), but they are fundamentally based on standard phantoms and scan parameters rather than true organ absorbed doses in individual patients. In children, the same CTDI_vol does not have equivalent dosimetric implications in infants, school-age children, and adolescents. Similarly, the same DLP may correspond to different organ exposures because of differences in scan range, body size, automatic tube current modulation (ATCM), and tissue attenuation. Therefore, pediatric CT requires a dose descriptor that more closely reflects patient-specific characteristics than CTDI_vol/DLP. Size-specific dose estimate (SSDE), proposed by the American Association of Physicists in Medicine (AAPM) Task Group 204 (TG-204), was developed to address the limitations of standard phantom-based dose indices that do not account for patient size. Subsequently, Task Group 220 (TG-220) introduced water-equivalent diameter (D_w), advancing size characterization from purely geometric dimensions to attenuation-corrected estimation; Task Group 293 (TG-293) further extended the SSDE methodology to head CT[4-6].

Accordingly, this review positions SSDE as an “intermediate bridging metric” in individualized pediatric CT dose assessment: Upstream, it is linked to scanner-output indices such as CTDI_vol/DLP; downstream, it connects with organ dose, effective dose, risk assessment, and image-quality optimization. This review mainly discusses why children require more accurate dose estimation, how SSDE improves upon conventional indices, how current evidence supports its clinical application, and how SSDE may be integrated with automated D_w calculation, organ-dose estimation, and task-based image-quality assessment in the future.

SPECIFIC FEATURES OF PEDIATRIC CT DOSE ESTIMATION: RADIATION SENSITIVITY AND INDIVIDUALIZED NEEDS

Radiation sensitivity and longer latency window in children

The specific requirements of pediatric CT dose estimation primarily arise from children’s greater biological sensitivity of pediatric organs and tissues to ionizing radiation. Children are in active growth and development; the hematopoietic system is active, bone marrow distribution is extensive, and the thyroid, breast, gonads, lens, and parts of the central nervous system remain under development or functional maturation. These tissues may therefore be more sensitive to radiation-induced injury. In a review of radiation risk from pediatric CT, Brody et al[7] stated that a small but nonzero carcinogenic risk from low-dose radiation should be assumed when children undergo ionizing radiation examinations. Brenner and Hall[8] also emphasized, from the perspective of increasing medical radiation exposure, that CT has become an important source of medical ionizing radiation exposure in the population. More importantly, pediatric radiation risk ultimately depends on the absorbed dose to specific organs, patient age, sex, tissue sensitivity, and irradiated volume rather than on a single scanner-output index such as CTDI_vol or DLP. Using pediatric phantoms on 64-slice CT, Feng et al[9] demonstrated that risk assessment in pediatric CT must consider examination site and organ distribution. Kutanzi et al[10] also emphasized that the longer follow-up window and biological sensitivity of children make dose optimization particularly important. Furthermore, because children have a longer life expectancy and a longer latency window for radiation-related long-term effects, the discussion of pediatric CT risk should avoid two extremes: Exaggerating risk and thereby delaying necessary examinations, and ignoring cumulative low-dose exposure and simply applying adult protocols. The population-based study by Dorfman et al[11] showed that ionizing radiation imaging procedures are not uncommon among children, making population-level dose management clinically meaningful. Westra[12,13] noted that CT risk communication should integrate age, exposed tissues, absorbed dose, and clinical benefit. Recent Asian population studies also support the importance of subsequent risk assessment and dose recording after pediatric CT: Wang et al[14] analyzed risks of leukemia, intracranial tumors, and lymphomas after CT exposure in childhood and early adulthood, and Han et al[15] further addressed subsequent cancer risk after pediatric CT in a nationwide population-based cohort study. Therefore, pediatric CT dose management should emphasize individualized and refined dose estimation on the basis of justification, avoid nonindicated examinations, repeated scanning, and excessive scan coverage, and adopt dose-assessment methods that reflect pediatric body size and organ-exposure differences.

Pediatric body-size variation and the limitations of CTDI_vol/DLP

Pediatric body size varies substantially from neonates and infants to school-age children and adolescents. Under the same scan protocol, smaller children may receive a higher absorbed dose per unit mass, and the clinical meaning of the same CTDI_vol differs across pediatric body sizes. Pages et al[16] conducted an early multicenter study of CT doses in children and evaluated weighted CT dose index, DLP, and effective dose in children aged 1 year, 5 years, and 10 years; their results showed that pediatric CT dose varied substantially by examination site, age group, and institutional protocol, suggesting the need for dose evaluation stratified by age and examination type. Hwang et al[17] surveyed pediatric CT protocols and doses in 19 Korean hospitals and showed that clinical scan parameters, CTDI_vol, DLP, effective dose, and SSDE varied with age and examination site. Their study also noted that in children younger than 5 years, SSDE correlated well with CTDI_vol, whereas in children aged 6 years or older, CTDI_vol was lower than SSDE, indicating that relying only on CTDI_vol may underestimate size-corrected dose in older children or children with different body sizes. Thus, pediatric dose assessment restricted to CTDI_vol/DLP may fail to detect dose mismatch caused by age, body size, scan coverage, and protocol differences.

The aim of the Image Gently campaign is not merely to reduce CT dose, but to tailor scanning parameters to pediatric body size and clinical task while maintaining diagnostic image quality. Strauss et al[18] proposed ten practical steps for pediatric CT optimization, including the use of “child-size” scan settings according to patient size, restriction of scan range, avoidance of unnecessary multiphase scanning, optimization of kilovoltage peak (kVp) and tube current-time product (mAs), and adjustment of image-quality requirements according to the diagnostic question. The International Commission on Radiological Protection (ICRP) Publication 121 further emphasized that pediatric diagnostic and interventional radiology should follow the principles of justification, optimization, dose recording, and age-appropriate radiation protection, indicating that pediatric imaging safety should be supported by institutionalized protocol management and continuous quality improvements rather than by isolated technical adjustments alone[19]. Within this framework, the value of SSDE lies in incorporating pediatric body size into CTDI_vol correction, thereby moving dose assessment from scanner-output monitoring toward patient-size-adapted dose management.

BASIC CONCEPTS AND METHODOLOGICAL EVOLUTION OF SSDE

Boundaries among CTDI_vol, DLP, effective dose, and SSDE

CTDI_vol is a volumetric dose index measured or derived under standard CT dose phantom conditions, and DLP is the product of CTDI_vol and scan length. These indices are suitable for scanner-output management and protocol comparison, but they do not represent organ absorbed doses in individual children. Effective dose is a tissue-weighted risk-related quantity that facilitates population-level comparison of radiation burden across examinations, but it is not appropriate for precise risk estimation in individual patients. Thomas and Wang[20] established age- and region-specific DLP-to-effective-dose conversion coefficients for pediatric multidetector-row CT examinations and showed that effective dose from a single-phase head CT decreased markedly with age: Approximately 4.2 mSv, 3.6 mSv, 2.4 mSv, 2.0 mSv, and 1.4 mSv in neonates and children aged 1 year, 5 years, 10 years, and 15 years, respectively. The study also noted a wide range of effective doses in pediatric CT, from < 1 mSv for ultra-low-dose protocols to 10-15 mSv for large-range body examinations, indicating that effective dose is substantially influenced by age, examination site, and scan range. Huda and Ogden[21] proposed a method for computing pediatric body CT effective dose using adult body CT effective dose, DLP, and ratios related to energy deposition; this approach indicates that DLP-based conversion to effective dose must consider age, body size, and examination site rather than directly applying adult conversion coefficients.

SSDE differs from effective dose. The basic form of SSDE is SSDE = CTDI_vol × size-conversion factor, with units of mGy, and it represents a size-corrected estimate of dose in the scanned region. Effective dose, by contrast, is expressed in mSv and represents a risk-related quantity weighted by tissue radiosensitivity. Newman et al[22] compared five methods for calculating effective dose in routine pediatric chest CT; their study assessed the strengths and limitations of different calculation approaches, demonstrating that pediatric CT effective-dose results depend on the selected algorithm, conversion coefficient, and dose model, and that the calculation basis must therefore be clearly reported. Brady et al[23] further noted that conventional DLP-based effective-dose estimation requires correction for patient size and proposed more appropriate effective-dose estimation using either SSDE or DLP. Their study aimed to explain how effective dose may be calculated using SSDE and how conventional DLP-only approaches can be modified. Therefore, in pediatric CT, CTDI_vol/DLP should be positioned as scanner-output indices, SSDE as a size-corrected patient dose estimate, and effective dose as a population-level risk-weighted quantity; these metrics may be related, but they should not be considered equivalent or used interchangeably.

TG-204: SSDE based on effective diameter

AAPM TG-204 is the foundational document for SSDE. It proposed that patient anteroposterior (AP) diameter, lateral (LAT) diameter, or effective diameter (D_eff) can be used to obtain a size-conversion factor for correcting CTDI_vol[24]. D_eff is typically calculated as the geometric mean of AP and LAT diameters and is convenient for clinical measurement and implementation. Brady and Kaufman[25] compared five pediatric CT SSDE calculation methods recommended by TG-204 and found that when AP diameter or LAT diameter was used alone, SSDE differences were within the range of 2%-12%, whereas when AP + LAT sum or D_eff was used, differences decreased to 0.9%-2%. Age-based estimation showed an average difference of approximately 2% in children aged 0-13 years, but in adolescents aged 14-18 years, the maximum difference reached 44% and the average difference was approximately 18%. These results suggest that actual size measurement, particularly AP + LAT or D_eff, should be prioritized in pediatric CT, rather than relying solely on age-based estimation.

The limitations of TG-204 are also clear: Its size metrics mainly reflect geometric size and do not fully capture tissue attenuation. Li and Behrman[26] commented on AAPM TG-204 and noted that the application and interpretation of size-conversion factors require attention to methodological assumptions[24]. In pediatric chest CT, the high air content of lung tissue creates a mismatch between geometric diameter and X-ray attenuation. In younger children and in children with marked differences in body composition, D_eff alone may also be insufficient for precise size characterization.

TG-220: From geometric size to D_w

The introduction of D_w by TG-220 represents an important methodological development in SSDE[6]. D_w integrates cross-sectional area and CT number information, thereby reflecting both patient geometry and tissue attenuation; it is therefore closer to the characteristics of X-ray energy deposition than D_eff. Anam et al[27] proposed an automated method for calculating D_w and SSDE, demonstrating the feasibility of automated D_w calculation from clinical CT images. Xu et al[28] compared SSDE calculated using D_eff and D_w in coronary CT angiography and found that the median SSDE_D_eff was 18.26 mGy, whereas the median SSDE_D_w was 20.56 mGy; SSDE_D_eff was approximately 10.08% lower on average, and the absolute relative difference increased linearly with the difference between D_eff and D_w. The fundamental reason for this difference is that D_eff reflects only geometric body size derived from AP and LAT diameters, whereas D_w incorporates cross-sectional area and CT number, thereby reflecting attenuation differences from lung, mediastinum, bone, fat, and other tissues. Thus, in chest-related examinations such as coronary CT angiography, D_eff and D_w do not represent equivalent patient size, and SSDE values calculated from them should not be used interchangeably.

In pediatric CT, D_w has even greater significance. The proportion of air-filled lung tissue in the chest, the distribution of soft tissue and fat in the abdomen, and the proportions of bone and organs in younger children all change with age, making pure geometric size insufficient for characterizing attenuation. Khan et al[29] analyzed the relationships among D_w, SSDE, and effective dose in pediatric head, chest, and abdominal CT and found that SSDE was significantly correlated with CTDI_vol and D_w (P < 0.05). This association has a clear dosimetric basis: CTDI_vol is the direct input for SSDE, so increasing CTDI_vol increases SSDE; D_w determines the size-conversion factor, such that at the same CTDI_vol, smaller children with lower D_w have larger conversion factors and relatively higher SSDE, whereas children with larger D_w have smaller conversion factors and relatively lower SSDE. The study also found a strong positive correlation between D_w and size/(LAT + AP), suggesting that diameter-based parameters may serve as alternative size metrics for SSDE estimation when automated D_w calculation is unavailable.

TG-293: A dedicated extension of SSDE to head CT

The SSDE framework for body CT cannot be directly applied to pediatric head CT (Table 1). Head CT is affected by skull attenuation, head circumference, lens position, scan angle, and age-related changes. AAPM TG-293 specifically proposed methods for SSDE calculation and conversion factors for head CT[4]. Jaramillo-Garzón et al[30] established age-stratified typical SSDE values for pediatric non-contrast head CT, with median SSDE values of 33.5 mGy, 31.6 mGy, 36.2 mGy, and 57.9 mGy for patients aged 0 months to < 3 months, 3 months to < 1 year, 1 year to < 6 years, and ≥ 6 years, respectively. Sapignoli et al[31] further stratified pediatric head CT into four D_w groups based on TG-293: < 14 cm, 14 cm to < 16 cm, 16 cm to < 17 cm, and ≥ 17 cm, and reported that using TG-204 body conversion factors for head SSDE could overestimate dose by up to approximately 12%, supporting the use of TG-293-specific conversion factors for pediatric head CT. Fahmi et al[32] compared different pediatric head CT protocols and found that the SSDE of the dedicated pediatric head protocol was markedly lower than that of routine axial and helical head protocols. However, the actual clinical use rates of the pediatric-specific protocol, routine axial protocol, and routine helical protocol were 20%, 37%, and 43%, respectively, suggesting that protocol selection itself is an important component of pediatric head CT dose optimization.

Table 1 Evolution of the size-specific dose estimate methodology and its application in pediatric computed tomography.

Methodology	Core parameters	Suitable scenarios for pediatric	Main contributions	Limitations
TG-204	AP, LAT, Deff, CTDIvol × body conversion factors	Pediatric body CT, especially used for basal dose correction of chest, abdomen, and pelvic CT	Easy to implement and traceable, the foundation of clinical popularization	Geometric dimensions cannot reflect tissue attenuation, and there may be deviations in chest and body composition scenarios
TG-220	Dw, cross-sectional area + CT value information	Scenarios with significant differences in chest, abdominal, and body composition among children	Closer to the attenuated body shape than Deff and represents a refined direction in SSDE development	Manual delineation time-consuming and poor repeatability, automatic delineation requires excellent algorithms
TG-293	Specialized conversion factor for head CT, head circumference/Dw-related stratification	Pediatric cranial CT examinations, non-enhanced cranial CT scans, and optimized cranial imaging protocols	Avoids inappropriate application of body CT conversion factors to head CT	Still cannot represent the dose received by organs such as the lens or thyroid gland
Automatic dimension calculation	Automatic boundary segmentation, batch calculation of Deff, Dw, SSDE	Dose database, DRL, continuous quality control, and abnormal dose alert system	Minimizing human error is crucial for clinical translation	Requires validation across scanners, protocols, vendors, and age groups

TG-204: Task Group 204; TG-220: Task Group 220; TG-293: Task Group 293; AP: Anteroposterior; LAT: Lateral; D_eff: Effective diameter; CTDI_vol: Conventional volume computed tomography dose index; D_w: Water-equivalent diameter; CT: Computed tomography; SSDE: Size-specific dose estimate; DRL: Diagnostic reference level.

SELECTION OF BODY-SIZE METRICS IN PEDIATRIC CT

Age and weight: Clinically practical but limited in precision

Age and body weight are the most readily available variables for pediatric CT protocol stratification. Age reflects developmental stage, but children of the same age may differ considerably in body size; weight is more closely related to body habitus than age, but it does not directly reflect cross-sectional attenuation or organ distribution. Khawaja et al[33] retrospectively analyzed 522 chest and abdominopelvic CT examinations in 483 children and found a good correlation between body weight and pediatric body diameter. They also showed that weight better reflected cross-sectional body-size variation than age, and therefore proposed that weight could be used as a substitute for body diameter to select SSDE conversion factors and simplify pediatric CT dose estimation. Kritsaneepaiboon et al[34] further validated this concept in 196 pediatric torso CT examinations. They found strong correlations between weight and each body-diameter parameter (r = 0.919-0.960, P < 0.001), and strong correlations between weight-based SSDE and SSDE calculated from AP diameter, LAT diameter, AP + LAT, and D_eff (r = 0.934-0.953, P < 0.001). The difference between weight-based SSDE and effective-diameter-based SSDE was the smallest. In addition, the median increase of weight-based SSDE over CTDI_vol was 88% (interquartile range: 66%-112%), suggesting that CTDI_vol underestimates size-corrected dose in most pediatric or small-body-size patients. Karmazyn et al[35] compared multiple patient-size surrogates and showed that, except for LAT diameter alone, geometric diameter- and weight-based approaches generally provided acceptable SSDE estimates when compared with D_w-based software calculation. More recently, Abdulkadir et al[36] validated a segmentation-based automated method for patient-size and SSDE calculation in pediatric CT, supporting the transition from manual measurement toward automated dose monitoring. Therefore, body weight is suitable for pre-scan protocol stratification and rapid dose estimation, but it should not replace image-based body diameter or D_w for precise SSDE calculation.

D_eff, AP/LAT diameter, and measurement standardization

D_eff, AP diameter, and LAT diameter are the most commonly used patient-size metrics in the TG-204 framework. Karmazyn et al[35] compared SSDE calculated from D_AP, D_LAT, D_AP + LAT, D_ED, and body weight with software-calculated SSDE based on mean D_w in 100 pediatric chest and abdominal CT examinations. They found that, except for D_LAT alone, most geometric metrics and the weight-based method provided acceptable estimates for pediatric body CT dose. The Bland-Altman 95% limits of agreement for the D_LAT method reached approximately 43%, whereas those for weight, D_AP, D_AP + LAT, and D_ED were generally within approximately 25%, indicating that single LAT diameter is more sensitive to measurement level and body-shape variation. Tsujiguchi et al[37], based on 753 CT examinations, further showed that SSDE conversion factors were closely related to patient size and that smaller patients may have higher size-corrected dose during a single scan, demonstrating that SSDE better reflects pediatric body-size differences than CTDI_vol. Therefore, pediatric SSDE studies should not merely report SSDE values; they should explicitly describe the source of size parameters, measurement level, use of scout or axial images, use of automated segmentation, and source of conversion factors, to ensure comparability across studies.

D_w and automated extraction of size metrics

D_w integrates cross-sectional area and CT number information and more closely reflects patient attenuation characteristics, but manual calculation is impractical for routine, large-scale dose monitoring. In 2024, Abdulkadir et al[36] validated a segmentation-based automated method for calculating patient size and SSDE in 4 CT dose index phantoms and 80 pediatric head, chest, and abdominal CT images. The results showed high agreement between the automated method and the AAPM manual method: Size estimation errors in patient images and phantom experiments were 1.9% and 0.27%, respectively; the percentage difference in SSDE was approximately 1%; and the Pearson correlations in clinical images and phantom experiments were r > 0.9771 and r > 0.9999, respectively. An earlier study by the same team showed that the mean D_eff values for head and abdominal CT in children aged 0-12 years were 14.79 cm and 16.33 cm, respectively, both smaller than reference phantom sizes. SSDE was significantly higher than scanner-displayed CTDI_vol (P < 0.001), with differences of 0%-17% for head CT and 37%-60% for abdominal CT, indicating that CTDI_vol may underestimate pediatric size-corrected dose[38]. Therefore, automated D_eff/D_w/SSDE calculation can reduce manual measurement error and support continuous dose quality control, pediatric dose database construction, and establishment of SSDE-based DRLs (Table 2).

Table 2 Comparison of computed tomography size-specific dose estimate body type parameters in pediatric.

Body size parameters	Data sources	Advantage	Limitations	Localization in pediatric CT
Age	Clinical data	Most readily available and convenient for initial protocol screening	Significant differences in body size among the same age pediatrics	Rough stratification variable
Weight	Clinical data	Available before scanning, related to body size	Does not reflect cross-sectional attenuation or organ distribution	Practical variable for protocol stratification
AP/LAT	CT topogram or cross-sectional image	Simple to measure and suitable for use with the TG-204	Influenced by slice and direction, it does not reflect organizational composition.	Basic geometric dimension specifications
Deff	AP and LAT geometric mean	Highly standardized, easy to promote	Differences in chest and body composition may result in potential biases	Common Metrics for Traditional SSDE
Dw	CT image area + CT value	Simultaneously reflecting size and attenuation	Requires reliable segmentation and validation across scanners, protocols, and body regions	Recommended indicators for the future
Automatic size segmentation	Image algorithms	High throughput, reproducible, suitable for quality control	Need to verify cross device generalization	Development direction of dose management system

TG-204: Task Group 204; CT: Computed tomography; AP: Anteroposterior; LAT: Lateral; D_eff: Effective diameter; SSDE: Size-specific dose estimate; D_w: Water-equivalent diameter.

APPLICATIONS OF SSDE IN DIFFERENT PEDIATRIC CT EXAMINATIONS

Pediatric chest CT

Chest CT is an important application area for pediatric SSDE research. Hopkins et al[39] found that the mean SSDE for body-size-adapted scans at a children’s hospital was lower than that observed for scans performed at community-site scans (8.68 mGy vs 13.29 mGy, P = 0.03), suggesting that SSDE can be used to evaluate whether clinical protocols are truly adapted to pediatric body size. Döwich et al[40] showed in 133 pediatric non-contrast chest CT examinations that, compared with SSDE based on D_w, CTDI_vol underestimated dose by 54.7%, 47.6%, 40.2%, and 31.2% in four chest effective-diameter groups (all P < 0.001). Compared with SSDE based on D_eff, CTDI_vol still underestimated dose by 47.6%, 39.4%, 27.0%, and 12.3% (all P < 0.001), demonstrating that patient size must be incorporated into pediatric chest CT dose assessment. Kim and Newman[41] applied a pediatric chest CT strategy based on weight-based reduction of kVp combined with low mAs. CTDI_vol, DLP, and effective dose decreased by 73%, 75%, and 73%, respectively, in children weighing < 15 kg, and by 45%, 44%, and 48%, respectively, in those weighing 15-60 kg. Although image noise increased by 55% and 41%, the examinations were still considered diagnostically acceptable[41]. Therefore, pediatric chest CT dose optimization should jointly consider SSDE, body size, image noise, and diagnostic task rather than pursuing reduction of CTDI_vol/DLP alone.

Pediatric abdominal and abdominopelvic CT

Pediatric abdominal and abdominopelvic CT generally involves a relatively long scan range and is frequently used in emergency care, trauma, appendicitis, and tumor evaluation; some examinations also involve contrast enhancement or multiphase scanning. Therefore, dose management cannot rely solely on CTDI_vol and DLP. Imai et al[42] calculated SSDE from pediatric AP and LAT diameters in 117 pediatric abdominal/pelvic CT examinations and used it to establish local DRLs at a Japanese national children’s hospital, indicating that SSDE can serve as an important supplement to CTDI_vol/DLP in pediatric abdominal/pelvic CT dose management. Hwang et al[43] further established local DRLs for pediatric abdominopelvic and chest CT based on body weight and D_eff. In abdominopelvic CT, CTDI_vol DRLs for the 5 kg, 15 kg, 30 kg, 50 kg, and 80 kg weight groups were 1.4 mGy, 2.2 mGy, 2.7 mGy, 4.0 mGy, and 4.7 mGy, respectively, whereas SSDE DRLs for D_eff groups of < 13 cm, 14-16 cm, 17-20 cm, 21-24 cm, and > 24 cm were 4.1 mGy, 5.0 mGy, 5.7 mGy, 7.1 mGy, and 7.2 mGy, respectively. SSDE was higher than CTDI_vol in all age groups, indicating that body-size stratification can more sensitively reflect pediatric dose differences. Suspected appendicitis in children illustrates the value of limiting scan length for dose optimization. Roberts et al[44] found that in 270 pediatric CT examinations in which the appendix was completely visualized, all appendices were located within 10.5 cm above the umbilicus. A height factor of 0.07 for focused CT identified 100% of appendices and reduced radiation exposure related to scan length by approximately 27% on average, whereas a height factor of 0.03 identified 97% of appendices and reduced exposure by approximately 43%. A subsequent study also showed that a height-based focused CT protocol reduced CT dose for suspected appendicitis without compromising diagnostic accuracy and used CTDI_vol and SSDE for adjusted analysis[45]. Therefore, pediatric abdominal/pelvic CT dose optimization should consider size-corrected dose, scan length, number of phases, and diagnostic task, rather than simply reducing mAs/kVp.

Pediatric head CT

Although head size varies less than body size, pediatric head CT dose estimation is still affected by head circumference, skull thickness, lens position, scan angle, and scan length. Fujii et al[46] investigated the relationship between SSDE and age or body weight in pediatric brain CT and showed that continuous age- or weight-based modeling can be used to estimate standard SSDE values and may help identify examinations with abnormally high dose. In addition, our previous head CT study applied the AAPM TG-293 framework and showed that head circumference can serve as a practical parameter for rapid absorbed-dose estimation during head CT[47]. Kamdem et al[48] established local DRLs for pediatric head CT. The 75^th percentile CTDI_vol/DLP values for ≤ 1 year, 1-5 years, 5-10 years, and 10-15 years were 28.6 mGy/545.8 mGy·cm, 32.6 mGy/735.0 mGy·cm, 37.1 mGy/761.6 mGy·cm, and 44.2 mGy/1081.2 mGy·cm, respectively, indicating that pediatric head CT still requires age-stratified and localized dose monitoring. In a pediatric brain CT/CT angiography dose study, Salah et al[49] reported CTDI_vol and DLP values of 40.9 ± 9.4 mGy and 866.1 ± 289.3 mGy·cm for non-contrast brain CT, and 40.6 ± 8.8 mGy and 850 ± 230 mGy·cm for contrast-enhanced brain CT, respectively, suggesting that pediatric brain CT dose remains an area that requires continued optimization. Therefore, pediatric head CT dose assessment should be based on TG-293 SSDE and further integrate scan angle, orbital avoidance, scan length, iterative reconstruction, and clinical indication. It should also be clearly stated that head SSDE cannot directly substitute for organ doses to the lens, thyroid, or brain.

Pediatric hybrid imaging and low-dose CT

In positron emission tomography/CT and single-photon emission CT/CT, the CT component is generally used for attenuation correction and anatomic localization, and its dose is generally lower than that of diagnostic CT; nevertheless, pediatric patients still require accurate dose recording and size-corrected evaluation. Rajaraman et al[50] studied the relationship between patient size, CTDI_vol, and SSDE under automatic exposure control in 111 myocardial perfusion SPECT/CT examinations. The mean D_eff was 26.2 cm, median CTDI_vol was 7.27 mGy, and mean SSDE was 10.6 mGy; CTDI_vol was positively correlated with D_eff (r = 0.536, P < 0.0005), while SSDE further incorporated the size-conversion factor and reflected dosimetric differences of CTDI_vol across different patient sizes[50]. Although this study was not pediatric-specific, it illustrates a general limitation of CTDI_vol under automatic exposure control and supports the conceptual relevance of SSDE in hybrid imaging. Low-dose CT must also be constrained by image quality. Raslau et al[51] found in a head CT dose-reduction study that dose reduction impaired gray-white matter differentiation, but increasing the strength of iterative reconstruction partially restored image quality. With an appropriate combination of dose and reconstruction strength, approximately 58% dose reduction still achieved image quality comparable to the reference images, and the authors concluded that approximately 60% dose reduction may be feasible while preserving diagnostic quality. Therefore, for children, SSDE may serve as a size-corrected dose constraint, whereas image noise, low-contrast resolution, lesion detectability, gray-white matter differentiation, or target anatomic-structure visibility should serve as quality constraints. The goal of pediatric low-dose CT optimization should not be dose reduction alone, but rather a balance between dose and diagnostic quality for the specific clinical task.

SSDE, ORGAN DOSE, AND RADIATION-RISK ASSESSMENT

Correlation between SSDE and organ dose

SSDE has the advantage of being rapidly derived from CTDI_vol and body-size parameters, which makes it suitable for clinical dose monitoring and large-sample quality control. However, pediatric radiation risk is more directly determined by organ dose than by a single scanned-region dose metric. Moore et al[52] used four anthropomorphic phantoms of different pediatric body sizes, placed metal-oxide-semiconductor field-effect transistor dosimeters at 23 organ locations, measured absolute organ doses, and derived organ dose-to-SSDE correlation factors. For organs entirely within the scan range, the mean organ dose/SSDE ratio was 1.1 (range 0.7-1.4) for chest CT and 0.9 (range 0.7-1.3) for abdominopelvic CT. Organ doses estimated from SSDE differed from previous Monte Carlo results by < 5% on average for chest CT and < 2% for abdominopelvic CT. However, for tissues extending beyond the scan range, such as skin, bone marrow, and bone surface, the correlation was markedly weaker, averaging approximately 0.3 (range 0.1-0.4). Franck et al[53] constructed patient-specific voxel models from whole-body CT data of 10 children aged 2-18 years and used Monte Carlo simulation to model dose distributions for chest and abdominopelvic CT. They found strong linear correlations between SSDE and organ dose (r > 0.8), and also between SSDE and both blood dose and lifetime attributable risk (both r > 0.9), suggesting that SSDE may provide a rapid approximation of organ dose and risk-related metrics in pediatric torso CT. Nevertheless, correlation does not imply replacement. Hardy et al[54] evaluated the representativeness of SSDE for routine head, chest, and abdominopelvic CT using Monte Carlo organ doses as the reference. Their study noted that AAPM TG-204 anticipated organ dose and SSDE to differ by approximately 10%-20%, but actual differences depend on examination site, organ location, scan range, patient size, and tube current modulation strategy. Therefore, in pediatric CT, SSDE should be described as a “size-corrected scanned-region dose estimate” or an “approximate predictor of organ dose”, rather than being equated directly with organ absorbed dose. For radiosensitive organs such as the lens, thyroid, breast, gonads, and bone marrow, especially when the organ is not fully included in the scan range or lies near the scan boundary, more accurate assessment should incorporate scan range, organ location, and, when needed, Monte Carlo or organ-dose models.

Monte Carlo, computational phantoms, and patient-specific organ dose

More accurate pediatric CT organ-dose estimation generally relies on age-specific computational phantoms and Monte Carlo simulation. Lee et al[55] performed organ-dose simulations for helical multislice CT using 10 pediatric computational phantoms and found that tube voltage substantially affected effective dose under the same scan conditions: Compared with 80 kVp, effective dose increased on average by 105% at 100 kVp and by 210% at 120 kVp. In addition, stylized and voxel phantoms produced systematic differences in effective-dose estimation for chest and abdominal CT; stylized phantoms tended to overestimate effective dose in chest CT and underestimate it in abdominal CT, indicating that phantom anatomy directly affects organ-dose estimation. The National Cancer Institute dosimetry system for CT (NCICT) further combines the ICRP adult and pediatric reference phantoms with Monte Carlo simulation, establishes organ absorbed dose/CTDI_vol dose coefficients, and supports tube-current modulation and batch calculation. Its validation showed that differences between NCICT and CT dose-page CTDI_vol were mostly < 12%, but could reach 20% in individual cases; if average mAs was used instead of modulated mAs, organ doses near the lung in chest and chest-abdomen-pelvis CT could be overestimated by up to 2.4-fold[56]. Phantom and Monte Carlo methods still require measurement validation. Dabin et al[57] validated calculation algorithms in a 5-year-old physical pediatric anthropomorphic phantom on five CT scanners from four manufacturers, measuring doses to 22 organs. They found substantial interscanner variation in measured organ doses, with a coefficient of variation up to 53% at 80 kV. After CTDI_vol normalization, however, the mean coefficient of variation decreased to 12%, and most measured and simulated organ-dose differences were within ± 20%, except for bone marrow, breast, and ovaries, suggesting that CTDI_vol-normalized Monte Carlo organ-dose estimation has acceptable reliability. Pan et al[58] developed a 1-year-old computational phantom and created an organ-dose database including 36 organs and tissues and 47 axial scans. Comparison between Monte Carlo calculations and helical-scan measurements in a 1-year-old physical phantom showed differences < 25% for all organs, supporting the use of age-specific phantoms in pediatric CT organ-dose assessment. Recently, organ-dose calculation has gradually progressed from research applications toward clinically usable tools. The NCICT 2.0 system developed by Lee et al[59] incorporated 351 adult and pediatric size-specific phantoms and tube-current modulation profiles; in validation with 10 abdominal CT patients, dose coefficients differed from program-calculated results by less than 13%. If only reference-size phantoms were used, organ dose in overweight patients could be overestimated by more than 80%, indicating that size-adaptive models can markedly reduce dose-estimation error. Therefore, SSDE is more appropriate as a size-corrected dose-screening metric in pediatric CT, while more accurate individualized dose assessment should integrate automated organ segmentation, age-/size-specific computational phantoms, and rapid Monte Carlo dose simulation.

Boundary between SSDE and risk assessment

Pediatric CT risk assessment cannot be directly replaced by SSDE, because risk models typically require organ dose, age, sex, tissue weighting factors, and epidemiological parameters. Chu et al[60] established DLP-to-effective-dose conversion coefficients using 128397 pediatric diagnostic CT examinations and computational phantoms. They showed that conversion coefficients decreased substantially with age and varied significantly by examination site. For example, the head CT conversion coefficient decreased from 0.039 mSv/mGy·cm in children aged < 1 year to 0.003 mSv/mGy·cm in those aged 15-21 years, an approximately 13-fold difference. The chest CT coefficient decreased from 0.285 mSv/mGy·cm to 0.042 mSv/mGy·cm, an approximately 6.8-fold difference. When stratified by body diameter, the chest CT conversion coefficient decreased from 0.171 mSv/mGy·cm in the 11-15 cm diameter group to 0.032 mSv/mGy·cm in the 31-35 cm group, an approximately 5.3-fold difference[60]. Furthermore, effective doses estimated using age- or diameter-based DLP conversion coefficients showed only moderate-to-strong correlations with Monte Carlo results, with correlation coefficients of approximately 0.52-0.80 and 0.60-0.80, respectively, indicating that even effective-dose estimation is substantially affected by age, body size, and examination site. Thus, although SSDE is superior to CTDI_vol/DLP because it incorporates pediatric body size and can assist with organ-dose approximation and risk stratification, it is not organ dose, effective dose, or cancer risk. Its optimal role is that of an intermediate metric in individualized pediatric CT dose management, linking scanner output, patient size correction, organ-dose estimation, and risk assessment.

APPLICATIONS OF SSDE IN PEDIATRIC CT DOSE MANAGEMENT

Pediatric DRLs

Pediatric DRLs should not simply adopt adult standards, nor should they be set only by broad anatomic region. Instead, they should incorporate age, weight, body diameter, or clinical indication whenever possible. Kanal et al[61], using the United States American College of Radiology Dose Index Registry, established national DRLs and achievable doses (ADs) for 10 common pediatric CT examinations, explicitly proposing that DRL/AD should be stratified by patient age and D_eff to guide institutions in adjusting protocols according to pediatric body size. A systematic review by Priyanka et al[62] showed substantial heterogeneity among pediatric CT DRL studies across countries and regions in age grouping, weight grouping, dose metrics, and examination categories, indicating that pediatric DRL standardization remains insufficient. Recent studies have further shown that pediatric DRLs are moving from simple anatomic stratification toward body-size, weight, and indication stratification. Bos et al[63] established indication-based pediatric CT DRLs using international registry data and noted that different clinical indications within the same anatomic region do not have the same dose requirements. They also indicated that in head CT, children older than 6 years still require further age subdivision, and that DRLs for routine-dose chest and abdominopelvic CT were approximately twice those for corresponding low-dose categories, suggesting that pediatric DRLs should reflect diagnostic task differences. Hwang et al[64] established local DRLs for pediatric neck CT at nine university hospitals in Korea, including 1159 examinations and stratifying by age, weight, and D_w. Weight-stratified CTDI_vol DRLs increased from 5.2 mGy in the < 10 kg group to 16.2 mGy in the ≥ 60 kg group, and age-stratified CTDI_vol DRLs increased from 5.3 mGy in the < 1 year group to 15.6 mGy in the ≥ 15 years group, indicating that even within pediatric neck CT, dose benchmarks vary significantly with body size. This study also showed marked interinstitutional variation: In the < 10 kg weight group, the maximum/minimum ratio of typical CTDI_vol values reached 5.8, indicating that local DRLs can identify protocol variation and dose outliers. Ploussi et al[65] also showed in Greek pediatric local DRLs that dose ranges differed widely by examination site and age/weight group: CTDI_vol LDRLs for head, chest, and abdominopelvic CT were approximately 15-65 mGy, 1-5 mGy, and 1-7 mGy, respectively, while DLP LDRLs were approximately 211-787 mGy·cm, 18-226 mGy·cm, and 36-438 mGy·cm, respectively. These data indicate that pediatric CT dose management requires localized, size-specific, and task-specific DRLs. In the future, SSDE can serve as a supplement to CTDI_vol/DLP for establishing SSDE-based DRLs that better reflect pediatric body-size differences, rather than evaluating pediatric scan dose only against adult standards or a single anatomic threshold.

Automated dose monitoring and cumulative dose management

Children with chronic diseases, tumors, complex congenital disorders, or recurrent abdominal pain may undergo multiple CT examinations; therefore, dose management should extend from single-examination recording to patient-level cumulative dose tracking. Simply summing of CTDI_vol or SSDE does not accurately reflect cumulative exposure, because each scan may differ in Z-axis coverage, tube-current modulation, scan center, and organ inclusion. Tabari et al[66] proposed patient-size-specific z-axis dose profiles and dose line integral (DLI) for patient-level CT dose monitoring. In children undergoing repeated abdominal/pelvic CT, they found that simple summation of SSDE values across repeated CT examinations underestimated the patient’s peak dose along the Z-axis. Their proposed DLI integrates size-corrected dose and scan length and is more suitable for describing cumulative dose distribution from repeated scanning. Therefore, pediatric dose management should not merely retain single-examination CTDI_vol, DLP, or SSDE values; it should also record scan start and end positions, scan length, z-axis dose distribution, and tube-current modulation information to identify local regions of repeated exposure. Boos et al[67] established institutional DRLs based on D_w and SSDE in 1690 chest, abdominopelvic, and upper abdominal CT examinations. Mean CTDI_vol, D_w, and SSDE were 7.2 ± 4.0 mGy, 29.0 ± 3.4 cm, and 8.5 ± 3.8 mGy, respectively, with SSDE higher than CTDI_vol. D_w and SSDE also showed a linear relationship, supporting size-specific DRL stratification by D_w. This framework suggests that pediatric automated dose platforms should simultaneously collect CTDI_vol, DLP, scan length, body-size parameters, D_w/D_eff, SSDE, kVp, tube-current modulation profiles, and reconstruction method, and should generate quality-control reports stratified by age, weight, examination site, clinical indication, and scan range. When an examination’s SSDE, DLI, or local cumulative dose is substantially exceeds than the institutional benchmark, radiologic technologists and medical physicists should be prompted to review the protocol.

Joint optimization of SSDE and image quality

Dose optimization is not simply dose reduction. A reasonable goal of pediatric CT is to use the lowest reasonable dose while satisfying the diagnostic task. Sayed and Mohd Yusof[68] reviewed techniques for minimizing radiation dose in pediatric abdominal CT and emphasized that children are more susceptible to radiation-induced cancer and have longer life expectancy; therefore, dose optimization should integrate indication selection, parameter adjustment, and alternative imaging modalities. SSDE itself contains no image-quality information. A low SSDE may correspond to inadequate image quality, whereas a high SSDE may be reasonable for a complex examination. Therefore, future studies should establish relationships between SSDE and image quality, image noise, contrast-to-noise ratio, low-contrast detectability, subjective quality scores, iterative reconstruction, or deep-learning reconstruction. Only by linking SSDE with the diagnostic task can true individualized optimization be achieved (Table 3).

Table 3 Summary of the translational roles of size-specific dose estimate in pediatric computed tomography dose management.

Application domain	Representative studies	Evidence summary	Practical implication
Pediatric DRL	Kanal et al[61], Hwang et al[64], Bos et al[63], Kamdem et al[48], and Ploussi et al[65]	DRL should be stratified by age, body weight, body diameter, examination site, and clinical indication	SSDE can complement CTDIvol and DLP for establishing pediatric DRLs that better reflect body-size differences
Automated dose monitoring	Bos et al[63], Abdulkadir et al[36], Anam et al[27]	Automated extraction of dose reports and automated calculation of Dw and SSDE can support high-throughput quality control	Become a core function of pediatric CT dose-management platforms
Patient-level cumulative dose assessment	Tabari et al[66]	Simple summation of CTDIvol or SSDE cannot characterize local peak dose along the Z-axis	Scan start and end positions, scan length, tube-current modulation, and dose line integral should be recorded
Image-quality constraints	Raslau et al[51], Kim and Newman[41], Sayed et al[68]	Dose reduction may alter image noise and low-contrast resolution; optimization should be based on the diagnostic task	SSDE should be jointly modeled with image-quality metrics to avoid diagnostically insufficient low-dose imaging

DRL: Diagnostic reference level; SSDE: Size-specific dose estimate; CTDI_vol: Conventional volume computed tomography dose index; D_w: Water-equivalent diameter; CT: Computed tomography; DLP: Dose-length product.

CURRENT LIMITATIONS AND FUTURE DIRECTIONS

Current research on pediatric CT SSDE still has several limitations. First, SSDE is a size-corrected dose estimate and is not organ dose or risk dose. Second, size metrics are not yet fully standardized; age, weight, D_eff, D_w, single-slice D_w, and scan-range-averaged D_w may yield different results. Third, head CT and body CT use different size-correction logic and should not share the same conversion system. Fourth, ATCM, scan range, and multiphase enhancement make dose distribution more complex, and a single SSDE value cannot characterize z-axis dose variation. Fifth, SSDE lacks an image-quality dimension and cannot answer whether the dose is sufficient to complete the diagnostic task.

Future studies should focus on five directions. First, pediatric-specific SSDE databases should be established and stratified by age, weight, D_w, examination site, and clinical indication. Second, automated D_w and SSDE calculation should be promoted to reduce manual measurement error. Third, SSDE should be integrated with automated organ segmentation, graphics processing unit-accelerated Monte Carlo simulation, and patient-specific organ-dose estimation. Fourth, task-based dose optimization should be developed by jointly modeling SSDE, image quality, and diagnostic task. Fifth, multicenter, cross-vendor, and cross-scanner studies are needed to validate the generalizability of pediatric SSDE-based DRLs. Ultimately, pediatric CT dose assessment should shift from “scanner-output dose recording” toward integrated optimization of “patient-specific dose, organ dose, image quality, diagnostic task”.

CONCLUSION

Pediatric CT dose estimation requires more refined patient-specific methods than adult CT because pediatric organs and tissues are more sensitive to ionizing radiation, body size and tissue composition vary more substantially, life expectancy is longer, providing a longer observation window for delayed effects. By correcting CTDI_vol using a size-conversion factor, SSDE provides an important bridge between conventional scanner-output indices and patient-specific dose assessment. With the development of D_w, TG-293, automated size measurement, and dose-management systems, SSDE is evolving from a simple dose-characterization tool toward a practical framework for pediatric CT protocol optimization, DRL establishment, and quality control. However, SSDE cannot replace organ dose or individual risk assessment. Future pediatric CT dose optimization should integrate SSDE with automated organ segmentation, Monte Carlo simulation, image-quality evaluation, and diagnostic task, thereby forming a truly patient-centered dose-management framework.

ACKNOWLEDGEMENTS

The author would like to thank the Department of Radiology team at Hubei Cancer Hospital for their previous contributions to computed tomography radiation dose estimation, optimization, and radiation protection.