Preamble

Title: Development Personalized Chinese Adult Phantom Library Accurate omputed omography Evaluation Authors:

Zhijie Zhang Xinliang Yuhang Tianliang Yongxian Zhang Dandan Yanjun Lihua Zhenchang antao Affiliations:

1 Department

Radiology, Beijing Friendship Hospital, Capital Medical University, Beijing China

2 School

of Physics, Beihang University, Beijing, 102206, China

3 Department

Radiology, Beijing Tongren Hospital, Capital Medical University, Beijing China

4 Department

Engineering Physics, Tsinghua University, Beijing, 100084, China address:

Abstract

Computed tomography indispensable clinical diagnosis.

Nevertheless, involves non-negligible risks exposure ionizing radiation. accurately evaluate radiation Chinese adults during examinations, analyze relationships between size, organ mass, anthropometric parameters Chinese adult population.

Anatomical contours obtained clinical images develop Chinese Adult Phantom (CAMP) library containing organs tissues.

Compared International Commission Radiological Protection esh-type eference omputational hantoms (MRCPs), exhibits markedly reduced relative errors organ masses. phantom, relative error liver decreased 67.38% 2.44%, spleen 47.35% 3.89%, pancreas 44.69% 8.08%. female phantom, relative error liver decreased 40.31% 9.53%, stomach 50.01% 10.35%, breast 70.77% 12.51%.

Compared Chinese national standard anatomical reference data, organ masses normal polygon model discrepancies within approximately Furthermore, organ masses morphologies phantoms representing different sizes remain anatomically realistic clinically consistent.

Ultimately, establish cohort phantoms accurately represent Chinese adults. present comprehensive system estimating organ doses examinations offer valuable insights optimizing imaging protocols assessing patient radiation exposure. words Computed tomography, monte carlo simulation, chinese adult phantom organ

1. Introduction

Currently, billion radiological diagnostic therapeutic procedures performed annually China contributing substantially patient healthcare while simultaneously posing potential risks associated ionizing radiation exposure 2016, approximately

million radiological diagnostic procedures conducted China, X-ray examinations accounting 52.4%, computed tomography scans 37.1%, interventional radiology 0.7%, nuclear medicine According 2020/2021 report United Nations Scientific Committee Effects Atomic Radiation (UNSCEAR), examinations contributed 61.6% collective effective radiological imaging procedures survey involving million patients across countries reveal nearly 900,000 patients worldwide receive cumulative radiation doses exceeding Furthermore, CT-related cancers could eventually account newly diagnosed cancer cases current utilization rates levels related study positive correlation between cumulative radiation exposure incidence hematologic malignancies Overall these findings underscore critical importance accurately quantifying radiation doses associated examinations. clinical examinations, current varies dynamically according patient size, volume computed tomography index reported scanner, which calculated using average current, cannot accurately represent absorbed specific organs.

Schlattl demonstrated organ coefficients derived fixed current conditions combined scanner-reported estimate organ doses, resulting errors exceed Monte Carlo simulation widely recognized accurate

method

radiation assessment plays pivotal fields nuclear engineering, radiation therapy, diagnostic radiology. evaluation, simulation estimates patient radiation exposure modeling energy deposition large number particles interacting human tissues.

Samei report organ reference values chest abdominal examinations under different scanning conditions using simulations However, substantial differences composition between Western Chinese adults, organ reference values derived Western models cannot directly applied Chinese patients.

Therefore, essential develop radiation assessment tools tailored Chinese adult characteristics establish reference clinical applications.

widely adopted technique radiation estimation diagnostic radiology, accuracy simulation largely depends computational phantom employed. development computational phantoms evolved through three major stages, including mathematical, voxel-based, mesh-based phantoms mathematical phantom represents organs tissues using simple geometric shapes ellipsoids, cones, elliptical cylinders. early limited computing power, these simplified anatomical models frequently employed simulations their simplicity, computational efficiency, adjustment, played essential establishment international radiation protection standards 1969, Snyder developed MIRD-5 (Medical Internal Radiation Dosimetry, MIRD) phantom represent average human anatomical characteristics However, because extensive anatomical simplifications, mathematical phantoms unable accurately capture complexity human anatomy. voxel phantom emerge advancement imaging technology.

These phantoms constructed stacking large number voxels, assigned unique identifier specific anatomical structure.

Unlike mathematical phantoms, voxel phantoms derived actual anatomical images, thereby providing higher degree anatomical realism.

Consequently, widely radiation assessment research. recent decades, extensive research efforts devoted development voxel phantoms. developed series voxel phantoms, including VIP-Man pregnant female model Bolch constructed series pediatric voxel phantoms Similarly, Zankl created series voxel phantoms based multiple individuals date, hundreds voxel phantoms developed worldwide, covering range size, groups, stages pregnancy Nevertheless, construction voxel phantoms constrained spatial resolution imaging systems.

Consequently, surface voxel phantom often exhibit stair-step appearance, modeling accuracy small anatomically complex structures remains relatively limited Mesh-type computational phantoms constructed using non-uniform rational B-splines (NURBS) polygon meshes represent organ. quality accuracy these phantoms depend density distribution mesh, enabling smoother surfaces detailed representation human anatomy. 2001, Segars developed first mesh-type

phantom modeling approach offers distinct advantages describing anatomical deformations 2005, introduced series phantoms, including RPI-AM, RPI-AF, RPI-pregnant female model developed Mesh-type Reference Computational Phantoms (MRCP), which published International Commission Radiological Protection Publication 2024, Publication introduced Paediatric Mesh-type Reference Computational Phantoms 20,21 addition, developed Chinese pediatric reference phantoms evaluating radiation doses pediatric examinations constructed series Chinese adult standing phantoms account variations developed library individualized pregnant female fetal phantoms, including phantoms spanning gestational weeks Chinese reference phantoms different individualized pediatric computational phantoms Although numerous studies constructed computational phantoms Chinese adults, existing individualized voxel models generated scaling standard reference models without accounting variations internal organs, external contours often anthropometric reference data. addition, these phantoms extensively applied clinical radiation protection therefore provided limited practical benefit clinical settings.

Consequently, developing highly individualized computational models Chinese adults essential improving accuracy radiation estimation. address these limitations, developed Chinese Adult Mesh-type Phantoms (CAMP) series, which incorporates shape characteristics organ properties representative Chinese adult population.

Furthermore, established conversion coefficients between clinical indices organ doses using self-developed radiation simulation program, providing practical reference rapid accurate estimation organ doses examinations.

2.1 Analysis

Individualized Characteristics Chinese Adults attenuation X-rays varies among patients different sizes, organ masses differ among adults varying physiques. establish Chinese Adult Mesh-type Phantom library comprising eight representative models reflect differences (underweight, normal weight, overweight, obese) (male female).

2.1.1 The

Relationship Between Organ Weight Strong positive correlations observed between weight masses heart, liver, spleen Similar weaker associations exist lungs kidneys, whereas brain independent weight strongly negatively correlated weight, organs heavier those females, except liver. report regression equations organ weight against weight based autopsy records collected anatomy, pathology, forensic departments major medical schools across China These analyses confirm linear correlation between organ weight, specific regression parameters presented Table Organs Female Brain Heart Kidneys Liver Pancreas Spleen quantify relationship between adult dimensions, height, weight Chinese population, utilize Human dimensions Chinese adults which presents fundamental statistics human dimensions circumference circumference circumference circumference circumference across different percentiles groups years) Linear regression performed obtain relation between human dimensions height weight regression coefficients listed Table

versus height weight Female Human imension study, polygonal primarily employed construct mesh-type phantoms, overall generation process illustrated First, organs tissues delineated images. accelerate process, TotalSegmentator algorithm applied achieve automatic segmentation individual organs tissues 31,32 images masks imported Slicer software visualization, where regions segmentation accuracy manually corrected re-delineated.

Subsequently, delineated organ contours exported models while preserving relative anatomical positions original images, these spatial relationships essential subsequent model integration. individual organ models further refined Rhinoceros software improve surface quality anatomical consistency. sizes organ models adjusted using offset commands deformation controllers ensure their masses match reference values.

Considering small organs skeletal structures minimal variation across adults different sizes, focus primarily constructing thoracic abdominal organ models. organs skeletal structures transplanted model further refined according specific requirements study.

After optimization, models assembled, their anatomical positions carefully refined based established anatomical knowledge.

Inevitably, overlapping geometries arose during assembly process, resolving these conflicts while

preserving anatomical plausibility presented substantial challenges. address issue, priority ranking assigned organ.

Dense organs those clinically important morphologically distinct features given higher priority preserved minimal modification.

Organ integration first performed using deformation controllers smooth organ surfaces maintain anatomical consistency. deformation alone insufficient eliminate overlap, Boolean subtraction operations applied separate intersecting regions.

Since Boolean subtraction create visible cross-sections reduce anatomical realism, applied minor overlaps limit artifacts. deformation subtraction prove insufficient, manual adjustment control points separate overlapping regions while maintaining organ masses consistent reference values.

These three strategies applied iteratively until inter-organ overlaps eliminated. hollow organs stomach, gallbladder, bladder, initial models exported Slicer represented solid structures which organ walls contents merged. differentiate these components, solid model defined organ wall, inner offset along average normal direction applied generate luminal volume.

After assembling complete human model, polygonal converted tetrahedral using POLY2TET program Workflow Constructing -type Phantom

2.2.2 Construction

Mesh-type Phantoms Chinese Adults Other

Rather merely scaling standard reference phantom, construct individualized mesh-type phantoms better capture anatomical variability. individualized phantom constructed independently following

method

described Section organ masses adjusted match reference values provided, which calculated based equation Table

1. Although

approach increases modeling workload, yields phantoms accurately represent physiological characteristics Chinese adults across different sizes.

Compared standard phantom, significant differences individualized phantoms observed adipose layers.

Therefore, mainly focus deforming tissue align phantom morphology individualized anthropometric dimensions. normal phantom initial model, around which deformation controller established.

Since occupies large volume, relatively number control points required achieve adjustments. study, number control points along respectively. control points regions upper chest, waist, thigh circumferences individually adjusted deform surface.

According previous studies, adipose tissue tends accumulate centripetally, concentrating trunk, particularly waist, abdomen, around navel, whereas female adipose tissue distributed chest, lower abdomen, distribution pattern taken account adjusting skin, dimensions iteratively calculated corrected until deformed matche reference measurements. tissue generated inwardly offsetting surface, followed further refinement ensure overall phantom consistent reference values.

2.3 Development

Radiation Simulation System computational model PHILIPS Spectral scanner developed construct radiation simulation system.

X-ray spectrum defined using UserSpectrum command GATE, spectra generated Spektr software reviously published scanner bowtie filter parameters

references

Additionally, following

method

proposed bowtie filter PHILIPS Spectral independently measured 39,42 resulting filter profile summarized

thickness thickness validate accuracy radiation simulation system developed study, organ doses first measured physical phantom under actual scanning conditions. voxel phantom constructed computationally, organ doses calculated using Monte Carlo simulations comparison.

ART-200A-5 phantom (manufactured PHILIPS Spectral scanner employed assess performance simulation system. simulated organ doses deviated measured values.

Considering additional uncertainties arising scout scans, material conversion, typical error levels reporte related studies, proposed

method

regarded capable accurately estimating organ doses during clinical examinations

2.4 Evaluation

Patient Organ Doses Routine Chest Abdominal Examinations study, chest abdominal scans subjects males females) collected investigate relationships among dose-length product size-specific estimates organ doses.

During automatic current modulation (ATCM), current varies along different axial positions patient.

Organ doses entire calculated summing products organ single axial slice current position. determine currents different axial locations, Python program implemented using Pydicom library extract current distributions original DICOM files.

Conversion factors between organ doses radiation metrics

derived fitting measured organ doses indices. These factors enable rapid estimation organ doses selected organs routine thoracoabdominal examinations.

Results

Personalized Phantom Library Chinese Adults images Chinese adults years collected, covering individuals different sizes primarily involving thoracic, abdominal, pelvic regions.

According Chinese classification standard, underweight define 18.5, normal weight overweight obesity Using aforementioned methodology, total eight individualized mesh-type phantoms representing Chinese adult males females different sizes constructed. models include underweight 16.73), normal 21.47), overweight 26.17), obese 32.00). female models include underweight 16.97), normal 21.83), overweight 25.34), obese 29.59). shows render images phantom library, while presents renderings after removal skin, tissue, muscle. shown Tables organ masses model exhibit substantial deviations Chinese national reference values, maximum relative errors 67.38% organs 70.77% female organs, further demonstrating necessity developing Chinese-specific computational phantoms. contrast, organ deviations models constructed study controlled within reference values. correcte discrepancies between organ masses Chinese reference value enabling accurate estimation organ effective doses Chinese adult patients.

Compared international reference phantom MRCP, models significantly reduced relative errors organ masses. phantom, relative error liver decreased 67.38% 2.44%, spleen 47.35% 3.89%, pancreas 44.69% 8.08%. female phantom, relative error liver decreased 40.31% 9.53%, stomach 50.01% 10.35%, breast 70.77% 12.51%. compared Chinese

national anatomical reference data, differences organ masses normal body-type within approximately Table compares organ weights individualized Chinese phantoms reference values.

Among these, brain weight showed significant variation across different sizes, while weights heart, kidneys, liver, lungs, pancreas, spleen increased weight.

Table presents anthropometric measurements individualized Chinese phantom models. series developed study includes eight models representing sizes males females, dimensions organ masses accurately reflect characteristics Chinese adults.

Compared phantoms recommended Publication phantoms exhibit smaller organ errors therefore suitable evaluating optimizing radiation doses Chinese adults.

However, study certain limitations. current models primarily focus thoracic abdominal organs although variations skeletal structures small organs relatively minor, their precise modeling requires further refinement future work. addition, number phantoms constructed remains limited, individuals different heights represented, which restricts general applicability model library. primary research establish highly accurate individualized mesh-type phantom population Chinese population integrate closely clinical applications optimize patient radiation exposure examinations.

Future studies focus developing phantoms representing various heights groups, particularly constructing Chinese pediatric mesh-type phantom series advancing optimization strategies.

Personalized Chinese Adult Mesh-type Phantoms

Chinese adult mesh-type phantom library tissue removed. right: underweight, normal-weight, overweight, obese sizes

Relative error Relative error Brain 3.93% 1.55% Adrenals 24.00% 5.64% Colon 7.44% 2.16% Small intestine 39.05% 3.89% Prostate reast

Relative error Relative error Brain 1.47% 0.99% Adrenals 18.97% 3.69% Colon 74.25% 9.35% Small intestine 67.96% 12.61% Uterus reast 70.77% 12.51%

underweight normal weight overweight obese Organs Female Female Female Female (1480.95) (1375.705) (1329.86) (1348.89) (454.51) (250.94) (283.98) (309.94) (346.52) (368.32) (241.24) (261.68) (277.74) (300.37) (1751.08) (1541.485) (1320.21) (1412.06) (1444.19) (888.89) (970.37) (1034.39) (1124.6) (138.81) (86.82) (109.22) (126.82) (151.62) (178.17) (151.64) (153.04) (154.14) (155.69) indicates reference values calculated equation table underweight normal weight overweight obese Human dimensions Female Female Female Female

3.2 Results

Simulations During scans, current varies across different axial slices, resulting varying

between organ doses Representative fitted curves conversion coefficients selected organs shown correlations between doses lungs, heart, liver, kidneys, pancreas, spleen 0.96, respectively.

Correlation

analysis

performed establish relationship between organ doses correlations between doses lungs, heart, liver, kidneys, pancreas, spleen respectively.

Correlation

analysis

performed establish relationship between organ doses correlations between doses lungs, heart, liver, kidneys, pancreas, spleen respectively. exhibits stronger correlation organ doses compared above

result

further demonstrate patient-specific metrics, which account individual habitus, provide accurate assessment organ levels. study further demonstrates patient-specific indices account individual characteristics accurately assess organ levels, consistent findings Franck linear relationship facilitates rapid clinical estimation organ doses examinations.

Future research build individualized Chinese adult mesh-type phantom population developed study further investigate personalized optimization strategies low-dose scanning. incorporating variations radiation absorption distribution among individuals different sizes, scanning parameters precisely adjusted minimize patient radiation exposure while maintaining image quality. addition, Monte Carlo based assessment

results

employed validate effectiveness safety individualized scanning protocols. contributes establishment low-dose optimization strategies tailored Chinese population provides solid theoretical technical foundation development personalized imaging protocols clinical practice, thereby enhancing safety precision examinations.

CTDIvol selected organs

4. Conclusion

study, library mesh-type phantoms reflecting personalized characteristics Chinese adults established, providing solid foundation radiation assessment supporting further clinical research related dose.

methodology

includes medical image acquisition, organ modeling optimization, model assembly, ultimately resulting cohort individualized phantoms representative Chinese adults.

Furthermore, examine relationships between organ weight, between anthropometric dimensions height weight, establish quantitative linear regression equations serve reference values constructing individualized phantoms.

results

indicate organ masses dimensions series mesh-type phantoms developed study better represent anatomical characteristics Chinese adults, making suitable evaluating optimizing radiation doses Chinese patients. addition, investigat relationships between organ doses clinical indices SSDE. experimental

results

exhibited stronger correlation organ doses. Overall, individualized Chinese adult phantom library developed study provides essential subsequent radiation assessments.

Acknowledgments: supported National Natural Science Foundation China U1867210 62227804) Beijing Scholar (Zhenchang Wang) Beijing Hospitals Authority Clinical medicine Development special funding support, code:

YGLX202501

References

1. Y. Zhang, Zhuo, necessity countermeasure strengthening radiation management diagnostic radiology, Chin.

Radiol. Prot. (2023). Liang, Study estimation medical exposure frequency China, Chin.

Radiol. Prot. (2019). United Nations Scientific Committee Effects Atomic Radiation, UNSCEAR 2020/2021 report.

Evaluation medical exposures ionizing radiation, York:

UNSCEAR (2022) Rehani, Yang, Melick, Patients undergoing recurrent scans: assessing magnitude, Radiol. (2020).

Smith-Bindman, Azman Firdaus, Projected Lifetime Cancer Risks Current Computed Tomography Imaging, Intern. (2025).

Bosch Basea Gomez, Thierry-Chef, Harbron, hematological malignancies radiation exposure children, adolescents young adults, (2023).

Schlattl, Zankl, Becker, conversion coefficients examinations adults automatic current modulation, Phys.

Biol. (2010).

1. E. Samei, Tian, database patient-based computational models benchmarking organ estimates Phys. (2020).

Zhao, Ning, Bingzhi, Review Computational Phantoms Quality Assurance Radiology Radiotherapy Deep-Learning 방사선방어학회지 (2022).

Zankl, Drexler, Petoussi-Henss, calculation external photon exposures using reference human phantoms Monte Carlo methods.

Organ doses parallel environmental exposure geometries (1997) Snyder, Ford, Warner, Estimates absorbed fractions monoenergetic photon sources uniformly distributed various organs heterogeneous phantom., Suppl (1969) Chao, Bozkurt, VIP-MAN:

Image-Based Whole-Body Adult Model Constructed Color Photographs Visible Human Project Multi-Particle Monte Carlo Calculations, Health Phys. (2000) Geyer, Reilly, UF/NCI family hybrid computational phantoms representing current population female children, adolescents, adults application dosimetry, Phys.

Biol. (2014). Petoussi-Henss, Zanki, Fill, family voxel phantoms, Phys.

Biol. (2002). exponential growth computational phantom research radiation protection, imaging, radiotherapy: review fifty-year history, Phys.

Biol. R233-302 (2014). Kainz, Neufeld, Bolch, Advances Computational Human Phantoms Their Applications Biomedical Engineering Topical Review, Trans.

Radiat. Plasma (2019). Segars, Tsui, Lalush, Development application dynamic Nurbs-based Cardiac-Torso (NCAT) phantom., Nucl. (2001) Zhang, Caracappa, RPI-AM RPI-AF, mesh-based, size-adjustable adult female computational phantoms using ICRP-89 parameters their calculations organ doses monoenergetic photon beams, Phys.

Biol. (2009). International Commission Radiological Protection, Publication Adult mesh-type reference computational phantoms, 49(3) (2020) Shin, Choi, Pediatric phantom library constructed mesh-type reference computational phantoms (MRCPs), Nucl.

Technol. (2024). Shin, Body-size-dependent pediatric phantom library constructed pediatric mesh-type reference compuational phantoms, Conf. 계학술발표회 (2022) Development Chinese mesh-type pediatric reference phantom series application assessment Chinese undergoing computed tomography scanning, Phys.

Biol. (2021). body-size-dependent series Chinese adult standing phantoms radiation dosimetry, Radiol.

Prot. (2023). Akhavanallaf, Zaidi, Construction patient-specific computational models organ estimation radiological imaging, Phys. (2019).

Huang, Chinese reference population: Open-source age-dependent computational phantoms reference Chinese population, Phys. (2025).

P.-A. Poletti, Platon, Assessment fetus pregnant female patient using patient-specific computational models, Radiol. (2018).

Grandmaison, Clairand, Durigon, Organ weight adult autopsies:

tables Caucasoid population, Forensic (2001). Bell, Long, Roden, Updating Normal Organ Weights Using Large Current Sample Database, Arch.

Pathol. (2022). Chen, Anatomical Physiological Metabolic Characteristics Chinese Reference (1998) Standardization Administration People Republic China, Human dimensions Chinese adults, (2023) Wasserthal, H.-C.

Breit, Meyer, TotalSegmentator: robust segmentation anatomical structures images, Radiol.

Artif. Intell. e230024 (2023). Antonoli, Berger, Indrakanti, TotalSegmentator Robust Sequence-independent Segmentation Multiple Anatomic Structures Radiology e241613 (2025).

Yeom, Choi, POLY2TET: computer program conversion computational human phantoms polygonal tetrahedral mesh, Radiol.

Prot. (2020). Research Deformation Technology Application Chinese Radiation Protection Reference (2014) Agostinelli, Allison, Amako, Geant4 simulation toolkit, Nucl.

Instrum.

Methods

Phys. Sect. Accel. Spectrometers Detect. Assoc. Equip. (2003).

Allison, Amako, Apostolakis, Geant4 developments applications, Trans.

Nucl. (2006). Sarrut, Bardi Advanced Monte Carlo simulations emission tomography imaging systems GATE, Phys.

Biol. 10TR03 (2021). Punnoose, Sisniega, Technical Note: spektr computational x-ray spectrum modeling analysis, Phys. (2016).

Yang, George Direct measurement filter profiles simultaneous geometrical calibration, Phys. (2017).

Zhang, technique characterize scanner bow-tie filter attenuation applications human cadaver dosimetry simulations, Phys. (2015).

Zhao, Schafer, indirect transmission measurement-based spectrum estimation

method

computed tomography, Phys. Biol. (2015). Yang, Ruan, filter profiles three modern scanners, Brief (2019).

Akhavanallaf, Zaidi, Assessment uncertainties associated Monte Carlo-based personalized dosimetry clinical examinations, Phys.

Biol. (2020). Franck, Vandevoorde, Goethals, Size-Specific Estimate (SSDE) patient-specific organ cancer estimation paediatric chest abdominopelvic examinations, Radiol. (2016).