A z-score-based reference system for renal volume using three-dimensional ultrasound: from model to clinical tool Authors: Wu Zhou, Ting Zhang, Hongyan Chen, Lichuan Yang, Yongzhong Li, Diming Cai, Li Qiu, Jiaojiao Zhou Published: 2026 ABSTRACT BackgroundAccurate assessment of renal volume is crucial for diagnosing nephropathies and evaluating living kidney donors. However, related studies have generally relied on linear models or focused on population-level associations, without providing clinically usable reference standards. This study aimed to develop a physiology-informed and clinically applicable reference system for renal volume through use of advanced nonlinear modeling.MethodsIn this cross-sectional observational study, 634 healthy living kidney donor candidates (aged 20–78 years) were prospectively enrolled at a tertiary medical center. Renal volume was measured with standardized three-dimensional ultrasound (3D-US). Associations with age, sex, body surface area (BSA), and laterality were modeled with a generalized additive mixed model (GAMM), which captured nonlinear relationships. Based on model predictions, standardized z-scores were derived, and clinically interpretable reference intervals were established. Nomograms and an online calculator were developed to facilitate real-world application.ResultsA total of 1,268 kidneys were analyzed [median age 51 years, interquartile range (IQR), 46–56 years; 67.0% female]. The GAMM explained 90.1% of the variance in renal volume. Renal volume showed a significant nonlinear association with age, peaking at approximately 53.5 years before declining. BSA was the strongest positive determinant. After adjustment, male sex (−7.93 mL) and right kidney laterality (−5.99 mL) were independently associated with lower renal volume. The derived z-score framework showed good calibration, with approximately 95% of observations within the reference range (−1.96 to +1.96). Nomograms and an online calculator enabled individualized estimation and interpretation of renal volume.ConclusionsThis study established a clinically interpretable reference system for renal volume by integrating 3D-US with nonlinear modeling. The z-score-based framework, supported by nomograms and an online calculator, enables individualized assessment, improving the clinical translation of renal volumetry. Introduction Total kidney volume (TKV) has evolved from a descriptive anatomical measure to a clinically meaningful surrogate of functional nephron mass, with direct implications for prognosis and renal outcomes. Beyond detecting congenital anomalies, accurate volumetric assessment informs risk stratification in patients with chronic kidney disease (CKD), predicts functional recovery following nephron-sparing surgery, and supports donor selection in kidney transplantation (1-3). Reduced kidney volume has also been shown to independently predict subsequent decline in estimated glomerular filtration rate (eGFR), even among individuals with mildly impaired renal function (4). Accordingly, interpreting individual measurements within robust population-based reference frameworks is essential for early risk identification and clinical decision-making (5). Although cross-sectional imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) are regarded as reference standards for renal volumetry due to their high spatial resolution and reproducibility, their use is often limited by cost, accessibility, and radiation exposure. As a result, there is a growing interest in alternative, more accessible methods. One promising technique is three-dimensional ultrasound (3D-US), in which the accessibility of ultrasound is combined with near cross-sectional imaging performance, making it a potentially valuable tool for renal volumetry. When applied with advanced imaging techniques, ultrasound offers a more affordable and less invasive alternative while still providing high-quality volumetric data. Advances in automated segmentation have further strengthened the associations between renal parenchymal volume and kidney function across a variety populations (6,7). These measurements are particularly relevant in specialized settings, including the monitoring of pediatric autosomal dominant polycystic kidney disease and the evaluation of living kidney donors, where volumetry may outperform conventional scintigraphic metrics (8,9). Large-scale population-based imaging studies have demonstrated that kidney volume is strongly associated with age, sex, and body size, supporting their incorporation into normative models (10). Despite its advantages—including its wide availability—two-dimensional ultrasound depends on geometric assumptions that systematically underestimate true renal volume, potentially compromising its accuracy in some cases (11). Moreover, studies examining this modality have typically employed simplified modeling approaches that inadequately capture nonlinear age-related patterns and fail to account for the intra-individual correlation between paired kidneys, potentially leading to biased estimates and reduced clinical interpretability (5,12). To address these limitations, we developed a z-score-based reference framework for renal volume in healthy adults using 3D-US, combining the accessibility of ultrasound with near cross-sectional imaging performance. A rigorously selected cohort of living kidney donors minimized confounding from subclinical disease. We further applied a generalized additive mixed model (GAMM) to model nonlinear age effects while accounting for intra-individual correlation and key determinants, including sex, body surface area (BSA), and laterality. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0011/rc). Methods Study design and population This cross-sectional observational study was conducted at West China Hospital of Sichuan University, with participants being recruited from the Departments of Medical Ultrasound and Urology. To ensure the integrity of a normative dataset, stringent exclusion criteria were applied. Individuals were excluded for congenital or acquired structural renal abnormalities (e.g., horseshoe kidney and prior unilateral nephrectomy), renal cysts or tumors larger than 3 cm in diameter, or any medical condition potentially altering renal morphology or compromising measurement accuracy. Specifically, individuals were deemed ineligible if they had uncontrolled diabetes mellitus (glycated hemoglobin >7%, reflecting suboptimal glycemic control based on commonly adopted clinical targets), thyroid dysfunction (defined as serum thyroid-stimulating hormone <0.4 or >4.0 mIU/L based on standard laboratory reference ranges, with or without abnormal free thyroxine levels), clinically significant blood pressure abnormalities (systolic blood pressure >160 or <90 mmHg) (13-15), a recent history (within 3 months) of abdominal surgery or urinary tract infection, or anatomical deformities such as severe scoliosis. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and was approved by the Institutional Ethics Committee of West China Hospital, Sichuan University [No. 2020(468)]. Informed consent was obtained from all participants. Data acquisition and ultrasound volumetry Demographic and anthropometric data were collected by trained research staff in accordance with standardized procedures. Height and weight were recorded to the nearest 0.1 cm and 0.1 kg, respectively, with participants wearing light clothing and no shoes. BSA was subsequently calculated via the Mosteller formula (16). All 3D-US examinations were performed by a single senior radiologist (>10 years’ experience in abdominal imaging) using the EPIQ 7/7C or EPIQ 5 ultrasound systems (Philips, Amsterdam, the Netherlands) equipped with broadband convex volume probes (frequency 1–5 MHz). During imaging, participants were instructed to hold their breath as a mechanical sweep captured the volumetric data of both kidneys. Acquired datasets were processed offline using QLAB software version 10.8 (Philips Ultrasound LLC, Bothell, WA, USA). For each kidney, manual delineation of the renal parenchymal boundaries was performed on eight equidistant planes, with the renal sinus fat being explicitly excluded (Figure 1); the final renal volume was automatically computed via the disk summation algorithm (17). Quality control included completeness and artifact checks by two independent radiologists. Additionally, volume measurements were subjected to a blinded review by a second senior radiologist. Any interobserver measurement discrepancies exceeding 5% were resolved in consensus meetings to ensure measurement reproducibility. 3D-US imaging and volumetric quantification workflow for renal assessment. (A) 2D grayscale ultrasound image of the right kidney, demonstrating clear visualization of the renal parenchyma and renal sinus, serving as the anatomical reference for subsequent 3D volumetric acquisition. (B) Two-dimensional grayscale ultrasound image of the left kidney, illustrating the characteristic morphology of the renal cortex, medulla, and renal sinus, enabling bilateral morphological comparison. (C1-C3) MPR views of the 3D-US volume dataset of the right kidney in coronal (C1), sagittal (C2), and axial (C3) planes, providing comprehensive spatial orientation of renal anatomy, facilitating accurate manual boundary delineation for volumetric analysis. (D1-D4) Key steps in 3D-US-based volumetric quantification of the left kidney: generation of eight equidistant sectional planes in QLAB 3D quantification software for guiding manual segmentation (D1), manual delineation of the renal parenchymal contour with exclusion of renal sinus fat (D2), 3D volume-rendered image providing an intuitive visualization of renal morphology (D3), and final renal volume output automatically calculated using the integrated disk summation algorithm (D4). 2D, two-dimensional; 3D, three-dimensional; MPR, multiplanar reconstruction; US, ultrasound. Statistical analysis All statistical analyses were conducted with R version 4.5.1 software (The R Foundation for Statistical Computing, Vienna, Austria). Data completeness was assessed prior to modeling. The baseline characteristics of the study population are summarized in Table 1. Missingness was minimal, with two missing observations for age and two for sex (each representing less than 1% of the cohort), and no missing values in renal volume, BSA, or laterality. Given the extremely low proportion and absence of discernible missingness patterns, simple imputation (mean for continuous variables and mode for categorical variables) was applied and is unlikely to have materially influenced model estimates. Continuous predictors (e.g., age and BSA) were further screened for spatial outliers to avoid distortion of nonlinear fits. Descriptive statistics summarized baseline cohort characteristics, and continuous variables are expressed as the median and interquartile range (IQR). The appropriateness of analytic approaches was verified by assessing variable associations via both mixed correlation matrices (linear relationships) and distance correlation matrices (computed with the energy package in R) for distance correlation. Table 1Descriptive statistics of the study variablesCharacteristicValue (N=634)Between-participant characteristic Age, years51 [46–56] Sex Female425 (67.03) Male209 (32.97) Body surface area, m21.64 [1.53, 1.77]Within-participant characteristic Renal laterality Left139.00 [124.10, 154.10] Right132.30 [117.10, 149.00]Participants were aged 20–78 years. Data are presented as median [Q1, Q3] or n (%). Descriptive statistics of the study variables Participants were aged 20–78 years. Data are presented as median [Q1, Q3] or n (%). Although the age distribution was concentrated in midlife—reflecting standard living donor eligibility patterns—the observed age range spanned 20 to 78 years. To mitigate potential instability in sparsely represented age strata, penalized smoothing within the GAMM framework was applied, and no extrapolation beyond the observed age range was performed. Determinants of renal volume were identified through GAMMs, implemented in R via the mgcv package. This flexible modeling framework allowed for specification of nonlinear effects of continuous predictors (through use of smooth splines) and accommodated the hierarchical data structure, as measurements on both kidneys were nested within participants. The model incorporated renal volume as the dependent variable; fixed effects included age, BSA, sex, and kidney laterality (left/right), while participant ID was modeled as a random effect to account for intra-individual correlation. Parameter estimation was conducted via the fast restricted maximum likelihood algorithm with discretization, optimized for computational efficiency given the sample size. A rigorous, data-driven procedure guided model selection, particularly regarding interaction terms. The ordered factor difference method was applied to formally assess the statistical significance of candidate interactions. Under this approach, the age-by-sex interaction was identified as significant (P=0.0336); however, the sex-by-BSA interaction was excluded after failing to reach statistical significance (P=0.2207). Model diagnostics were performed via the gratia package in R and comprised residual assessment for normality and homoscedasticity, ensuring the validity of model assumptions. Statistical tests were considered significant at a two-sided P value <0.05. Development of a clinically interpretable reference framework To translate model-based estimates into clinically applicable reference standards, we developed a standardized framework based on z-scores derived from the final GAMM. Calculation of z-score For each kidney, the z-score was calculated as follows: z = (observed renal volume − predicted renal volume)/residual standard deviation Where the predicted renal volume was obtained from the fitted GAMM incorporating age, BSA, sex, and renal laterality, while the residual standard deviation was derived from the model scale parameter. This standardization enabled direct comparison of individual measurements against the expected physiological distribution. Definition of reference intervals Clinically interpretable reference ranges were defined based on the z-score distribution. A z-score between −1.96 and +1.96 was considered within the normal range (corresponding approximately to the 2.5th–97.5th percentile interval), while values outside this range indicated abnormal renal volume. Nomogram development To enhance bedside usability, faceted nomograms were constructed, stratified by sex and renal laterality. These nomograms provide a graphical representation of predicted renal volume across age and BSA, allowing clinicians to visually estimate expected values and deviations. Clinical interpretation and decision framework A structured clinical interpretation system was established by linking z-score intervals to clinically meaningful categories, including renal atrophy, normal variation, and hypertrophy. Additionally, a decision-support workflow (Figure 2) was developed to guide clinicians through the sequential process of z-score calculation, interpretation, and corresponding clinical decision-making. Clinical decision pathway for interpreting kidney volume z-scores. This figure illustrates the clinical decision pathway for interpreting kidney volume z-scores based on the difference between observed and predicted renal volumes. The z-score is calculated by dividing the difference by the residual SD. The pathway provides guidance on further diagnostic steps and clinical interventions based on the severity of the deviation, ranging from routine monitoring to further diagnostic work-up for significant kidney abnormalities. BSA, body surface area; SD, standard deviation. Online calculator implementation To further facilitate real-world application, an interactive web-based calculator was implemented as a user-friendly application. This tool allows users to input patient-specific parameters (age, sex, BSA, renal laterality, and measured renal volume) and automatically outputs the predicted renal volume, z-score, corresponding clinical interpretation, and suggested diagnostic classification. The application is publicly available online (https://nice88099.github.io/renal-volume-calculator; accessed on March 23, 2026). The source code is hosted on a public GitHub repository, ensuring transparency and reproducibility. This implementation enhances accessibility and enables immediate clinical application of the proposed reference framework. Results Study population and preliminary analysis The selection process for the analytic cohort is depicted in the flowchart of participant inclusion (Figure 3). Of the 647 individuals initially screened as potential living kidney donors, 634 participants met all eligibility criteria and passed data quality assessments, contributing a total of 1,268 kidney observations to the final analysis (Figure 3). Flowchart of participant selection and exclusion criteria for the 3D-US renal volume study. The diagram illustrates the sequential screening process for establishing the final analytical cohort. Out of 647 initial outpatients undergoing 3D-US examination between January and August 2022, 634 participants (contributing a total of 1,268 kidneys) were included in the final analysis. Exclusions were applied hierarchically for structural renal abnormalities, clinical conditions potentially influencing renal volume (e.g., poorly controlled diabetes and severe hypertension), and data quality issues (extreme outliers detected during preprocessing). This rigorous selection process ensured the derivation of reference framework from a strictly defined healthy population. 3D, three-dimensional; US, ultrasound. To assess the structure of the dataset prior to model fitting, correlation analyses were performed. Heatmaps of the correlation matrices revealed that BSA exhibited the strongest positive linear correlation with kidney volume (mixed r=0.53). By contrast, the linear association between age and kidney volume was negligible (r=−0.02); however, distance correlation analysis identified a meaningful nonlinear dependency between age and kidney volume (r=0.08) (Figure S1). Visual exploration supported these findings: scatter and box plots (Figure S2) illustrated a clear linear trend between BSA and kidney volume, whereas the relationship with age showed a curvilinear pattern. Notably, BSA distributions varied by sex, underscoring the importance of multivariable adjustment in subsequent analyses. Determinants of kidney volume: primary model outcomes A GAMM was constructed to estimate the independent effects of key predictors on kidney volume while accommodating the within-participant correlation of paired kidneys. The model demonstrated excellent explanatory capacity, accounting for approximately 90.1% of the variance in kidney volume (adjusted R2=0.901). Parameter estimates for the fixed effects are provided in the summary table of the model (Table 2). Table 2Parameter estimates of the GAMMComponentTermEstimate/edfStd error/Ref.dft/F valueP valueA. Parametric coefficientIntercept139.24†1.75§t=79.72<0.001***Renal LateralityRight−5.99†0.39§t=−15.56<0.001***SexMale−7.93†2.51§t=−3.160.002**B. Smooth termss(Age)2.64‡2.72¶F=2.660.03*s(Body Surface Area)1.79‡1.83¶F=153.85<0.001***s(Age):SexFemale2.81‡2.87¶F=1.610.26s(Age):SexMale1.09‡1.09¶F=0.020.93s(ID)573.09‡630.00¶F=11.10<0.001***Adjusted R2 value: 0.901; deviance explained: 94.7% of the total deviance. Fast restricted maximum likelihood: 5,029.364; Scale estimate (residual variance): 46.996. s(Age) represents the smoothing function for age; s(Body Surface Area) represents the smoothing function for body surface area; s(Age):SexFemale and s(Age):SexMale represent the age-by-sex interaction smooths; s(ID) represents the participant-specific random intercept; edf denotes estimated degrees of freedom; Ref.df denotes reference degrees of freedom. Significance levels: ***, P<0.001; **, P<0.01; *, P<0.05. †, estimate; ‡, edf; §, Std error; ¶, Ref.df. GAMM, generalized additive mixed model. Parameter estimates of the GAMM Adjusted R2 value: 0.901; deviance explained: 94.7% of the total deviance. Fast restricted maximum likelihood: 5,029.364; Scale estimate (residual variance): 46.996. s(Age) represents the smoothing function for age; s(Body Surface Area) represents the smoothing function for body surface area; s(Age):SexFemale and s(Age):SexMale represent the age-by-sex interaction smooths; s(ID) represents the participant-specific random intercept; edf denotes estimated degrees of freedom; Ref.df denotes reference degrees of freedom. Significance levels: ***, P<0.001; **, P<0.01; *, P<0.05. †, estimate; ‡, edf; §, Std error; ¶, Ref.df. GAMM, generalized additive mixed model. Consistent with the exploratory analyses, BSA emerged as the most influential determinant, exhibiting a strong positive association with kidney volume (smooth term P<0.001). Renal laterality showed a significant asymmetry, with right kidneys being smaller than left kidneys by an adjusted mean difference of 5.99 mL (P<0.001). After BSA and age were controlled for, male sex was independently associated with a lower kidney volume compared to female sex (−7.93 mL; P=0.002), suggesting residual sex-specific biological effects beyond body size. Nonlinear age-related patterns and interaction analyses The GAMM-derived smooth terms revealed a significant nonlinear association between age and kidney volume. Partial effect plots demonstrated an inverted U-shaped pattern, with kidney volume increasing in early adulthood, reaching a peak at approximately 53.5 years, and subsequently declining (smooth term P=0.03) (Figure 4). This finding confirms that age-related changes in renal volume cannot be adequately captured through linear modeling approaches. Nonlinear partial effects of demographic and anthropometric determinants on kidney volume derived from the GAMM. (A) The overall smoothing curve for age. The trajectory identifies a specific physiological peak at coordinates (53.54, 7.26), indicating that kidney volume is maintained or slightly increases until approximately 53.5 years of age before declining. (B) Age-related trends in kidney volume stratified by sex (blue: male; red: female). (C) The partial effect of BSA, showing a steep, near-linear positive correlation with kidney volume. (D) Quantile-quantile plot for the random intercept s(ID), confirming the normality of random effects. Points represent individual data, lines represent GAMM-smoothed trends, and the shaded areas represent 95% confidence intervals. BSA, body surface area; GAMM, generalized additive mixed model; s(ID), participant-specific random intercept. Interaction analyses were conducted to assess potential effect modification. In the fully adjusted GAMM, the smooth main effect of age was not statistically significant within either sex (female: P=0.26; male: P=0.93). However, the age-by-sex interaction estimated with the ordered-factor difference approach was statistically significant, indicating a significant difference in the age-related kidney volume pattern between males and females (Table S1). In contrast, the interaction between sex and BSA was not significant (P=0.2207), suggesting that the association between body size and kidney volume was similar across sexes. Model diagnostics supported adequate model fit, with residual analyses indicating no major departure from normality or homoscedasticity (Figure S3). Development of a clinically interpretable reference framework for z-score distribution and reference intervals Based on the final GAMM, individualized predicted renal volumes were obtained for each observation, and standardized z-scores were calculated. The distribution of z-scores approximated a normal distribution centered around zero, indicating good model calibration and absence of systematic bias. Under this standardized framework, clinically applicable reference intervals were established. Approximately 95% of observations fell within the range of −1.96 to +1.96, supporting this interval as the normal reference range for renal volume. Values outside this range indicated abnormal renal size. Nomogram-based visualization of the reference framework Faceted nomograms were developed to visualize the joint effects of age and BSA on renal volume, stratified by sex and renal laterality (Figure 5). These graphical tools provide an intuitive method for estimating expected renal volume and identifying deviations from physiological norms in individual patients. Faceted nomograms of predicted renal volume by age stratified by sex, laterality, and body surface area. The panels display model-derived predicted renal volume (on three-dimensional ultrasound) as a function of age for the left and right kidneys, separately in females and males. Colored curves represent different BSA levels (1.27, 1.48, 1.69, 1.90, and 2.12 m2), while shaded bands indicate the corresponding 95% confidence intervals. Renal volume demonstrates a nonlinear association with age, characterized by an increase in early adulthood, a peak in midlife, and a subsequent decline. Across all strata, higher BSA is consistently associated with larger renal volume. The right kidney shows systematically lower predicted volumes compared to the left, and sex-specific differences are evident after adjustment for BSA. These nomograms provide a visual reference for estimating expected renal volume and support individualized clinical assessment. 3D, three-dimensional; BSA, body surface area. Clinical interpretation framework To translate statistical outputs into clinically meaningful categories, a z-score interpretation table was constructed (Table 3). This framework links quantitative deviations to clinical classifications, including renal atrophy, normal variation, and hypertrophy, thereby facilitating standardized interpretation in clinical practice. Table 3Formula for z-score calculation and clinical interpretation of three-dimensional kidney volumez-score rangeInterpretationClinical complication≤−1.96Severe deviation from normal, suggesting significant structural decline, with potential clinical concern for nephropathyPatients with a z-score <−1.96 may require further diagnostic testing, including imaging, to assess the degree of renal dysfunction or pathology−1.961.96Extreme deviation from the normal range, potentially indicating significant hypertrophy or pathological changesPatients with z>1.96 may need further diagnostic work-up to clarify the possible causes of abnormal renal growth, including conditions such as obesity or renal disease Formula for z-score calculation and clinical interpretation of three-dimensional kidney volume Clinical workflow for decision support A structured clinical workflow integrating z-score calculation, interpretation, and decision-making is presented in Figure 2. This framework outlines a stepwise pathway from raw volumetric measurement to clinical classification, supporting consistent and reproducible application in routine practice. Implementation of the online calculator An interactive web-based calculator was developed to enable real-time clinical application. By inputting patient-specific variables (age, sex, BSA, renal laterality, and measured renal volume), users can obtain predicted renal volume, z-score, corresponding interpretation, and suggested clinical classification. The outputs from the calculator were consistent with model-derived estimates, demonstrating its reliability and practical utility. The online calculator is publicly available and enables real-time individualized assessment, with outputs consistent with model-derived estimates. Discussion Principal findings This study established a clinically applicable reference framework for renal volume based on a rigorously screened cohort of healthy living kidney donors. By integrating GAMM-derived predictions with a standardized z-score system, we translated complex nonlinear modeling into an interpretable and clinically actionable tool. Three principal findings emerged in the study. First, renal volume exhibited a significant nonlinear cross-sectional association with age, characterized by an inverted U-shaped trajectory with a population-level peak at approximately 53.5 years (+7.26 units partial effect; P=0.03). Second, BSA was confirmed as the dominant determinant of renal volume (P<0.001), while the independent effects of sex and laterality persisted after adjustment, with male sex associated with smaller volumes (−7.93 mL; P=0.002) and right kidneys being smaller than left kidneys (−5.99 mL; P<0.001). Third, the model demonstrated excellent explanatory performance (adjusted R2=0.901), supporting its robustness as a foundation for deriving reference standards. Importantly, beyond identifying determinants, this study provides a complete clinical translation pathway, including z-score-based reference intervals, nomogram visualization, a structured interpretation framework, and an online calculator, thereby bridging the gap between statistical modeling and bedside application. Comparison with previous studies From population associations to physiological reference standards Large-scale imaging studies, including the German National Cohort (~10,000 adults) (6) and longitudinal CT-based analyses (5), have established population-level associations between renal volume and anthropometric and clinical variables. Additional population-based studies have further demonstrated significant associations between MRI-derived kidney volume and renal function (10), while reduced kidney volume in CT-based cohorts has been linked to subsequent renal function decline (4). However, these cohorts typically include individuals with subclinical or overt disease, which may bias reference estimation. In contrast, our study derived a reference framework from a rigorously screened living donor population, providing a physiologically grounded baseline with minimal pathological confounding. Despite differences in imaging modality (3D-US versus MRI/CT), previous studies consistently identified BSA as the primary determinant of renal volume. Ultrasonography also demonstrates broad applicability in evaluating retroperitoneal structures, supporting its role in renal imaging across diverse clinical scenarios (18). Re-examining the nonlinear age association Previous studies have often modeled renal volume with simplified, typically linear assumptions across age, particularly in heterogeneous or disease-inclusive cohorts. However, evidence suggests that renal structural changes with aging are nonlinear, characterized by progressive nephron loss, cortical thinning, and increasing nephrosclerosis (19-21). Our findings extend these observations by demonstrating an inverted U-shaped trajectory, with a midlife peak followed by decline. This pattern is physiologically plausible, as compensatory hypertrophy in early adulthood may temporarily offset nephron loss, whereas structural degeneration predominates beyond midlife. Importantly, given the cross-sectional design, the observed pattern reflects between-individual differences rather than longitudinal intra-individual changes. Nevertheless, it provides a biologically informed reference distribution that may improve interpretation of age-related deviations in renal volume. Sex and laterality effects in context After adjustment for BSA, male kidneys were found to be smaller than female kidneys in this cohort. Although this finding contrasts with some MRI- and CT-based studies, sex-related differences in renal size have been inconsistently reported and may depend on population characteristics and adjustment strategies (22-24). Notably, large population-based imaging studies have generally reported larger renal volumes in males, even after accounting for body size (23), whereas studies in carefully selected donor populations or alternative imaging modalities have shown attenuated or absent sex effects (25), highlighting the influence of cohort composition and methodological heterogeneity. Consistent with previous anatomical and imaging studies, we confirmed that the right kidney is smaller than the left (22,23). This asymmetry is generally attributed to anatomical constraints imposed by the liver and differences in vascular configuration. This is often overlooked in clinical practice, and our modeling quantifies this laterality effect after adjustment for confounders, supporting its inclusion in reference standards. From model to clinical tool: defining “normal” A key contribution of this study lies in transforming statistical outputs into a clinically interpretable reference framework. Studies in this area have largely reported regression coefficients or partial effect plots, which are difficult to translate into routine practice. In contrast, we implemented a z-score-based framework, analogous to established clinical standards (e.g., growth charts), enabling direct classification of individual measurements relative to expected physiological norms. By defining reference intervals (−1.96 to +1.96), we provide a standardized approach for distinguishing normal variation from pathological deviation, allowing consistent and reproducible bedside interpretation. Importantly, the integration of nomograms, interpretation tables, and a structured clinical workflow renders this framework operational rather than descriptive. The accompanying online calculator further enables real-time, individualized assessment. Opportunistic screening potential of 3D-US Our findings support an emerging role for 3D-US in opportunistic screening. Previous studies on diabetic populations have suggested that volumetric imaging can detect the early structural alterations associated with renal dysfunction (26). A growing body of evidence has linked renal volume to functional outcomes. MRI-based volumetry correlates with postdonation eGFR (9), while population-based CT studies have demonstrated that reduced kidney volume is associated with a subsequent decline in renal function (4). In addition, opportunistic CT-based analyses incorporating kidney volume and perirenal fat have shown potential for the early detection of CKD in routine imaging settings (27). Together, these findings support renal volumetry as a surrogate marker of nephron mass and functional reserve. The validity of ultrasound-based volumetric assessment is supported by studies demonstrating good agreement between 3D-US and MRI-derived kidney volume measurements (11). The integration of 3D-US into routine abdominal imaging workflows allows for the extraction of volumetric data without additional radiation exposure or substantial cost (28), supporting its feasibility for opportunistic use. This approach aligns with the broader paradigm of opportunistic imaging established in CT-based research (29) and further offers key advantages, including accessibility and the absence of ionizing radiation. Recent research further suggests that adiposity-related imaging markers, particularly perirenal fat, are closely associated with CKD risk (30), reinforcing the biological plausibility of integrating structural and metabolic imaging features into opportunistic screening strategies. Importantly, when combined with a z-score–based reference framework, volumetric assessment enables individualized interpretation relative to expected physiological norms. From a translational perspective, deviations from age-expected and body size-expected renal volume may indicate early structural vulnerability preceding measurable functional decline, supporting the role of renal volume measurement in early screening and risk stratification. Mechanisms and interpretation The observed nonlinear age pattern likely reflects a dynamic balance between compensatory hypertrophy and progressive nephron loss at the population level. During early adulthood, adaptive structural responses may partially offset nephron reduction, whereas beyond midlife, structural degeneration—driven by glomerulosclerosis, cellular senescence, and fibrosis—predominates (19-21). Importantly, these patterns reflect cross-sectional age-related differences rather than intra-individual longitudinal changes. Beyond BSA, renal structure is influenced by metabolic and adiposity-related factors. Renal sinus fat and perirenal fat have been linked to alterations in renal morphology and intrarenal hemodynamics (31-33), suggesting that body size alone does not fully account for variation in renal volume and that metabolic context is a highly influential factor. Persistent sex differences after BSA adjustment may reflect hormonal influences or differences in nephron endowment (34,35), underscoring the complexity of biological scaling in renal anatomy and its dependence on the population and physiological context. Strengths and limitations This study possessed several notable strengths. First, the use of a rigorously screened donor population minimized confounding from overt and subclinical disease, enabling derivation of a physiologically grounded reference framework. Second, the application of GAMM allowed for the robust modeling of nonlinear relationships and hierarchical data structures. Third, and most importantly, this study translated statistical modeling into a clinically applicable system, including z-scores, nomograms, and a deployable calculator. Several limitations should also be acknowledged. The cross-sectional design precludes inference regarding intra-individual longitudinal changes. The age distribution is concentrated in midlife, which may limit precision at age extremes. In addition, the single-center design and cohort composition may limit generalizability, and the findings should be interpreted as population-specific rather than universally applicable. Finally, although 3D-US is accessible and radiation-free, it remains operator-dependent compared with automated CT/MRI segmentation techniques. Prospective longitudinal studies and multicenter, multiethnic validation are required to further establish clinical utility and generalizability. Conclusions This study defined a clinically interpretable reference framework for renal volume in a rigorously characterized healthy population. By integrating nonlinear modeling with z-score standardization, it enables the individualized assessment of renal size within a cross-sectional context. Deviations from expected volume may indicate early structural vulnerability preceding functional decline. Prospective longitudinal validation is required to confirm clinical utility and generalizability. Supplementary The article’s supplementary files as