1
|
Huang Y, Holcombe SA, Wang SC, Tang J. A deep learning-based pipeline for developing multi-rib shape generative model with populational percentiles or anthropometrics as predictors. Comput Med Imaging Graph 2024; 115:102388. [PMID: 38692200 DOI: 10.1016/j.compmedimag.2024.102388] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/08/2023] [Revised: 04/06/2024] [Accepted: 04/18/2024] [Indexed: 05/03/2024]
Abstract
Rib cross-sectional shapes (characterized by the outer contour and cortical bone thickness) affect the rib mechanical response under impact loading, thereby influence the rib injury pattern and risk. A statistical description of the rib shapes or their correlations to anthropometrics is a prerequisite to the development of numerical human body models representing target demographics. Variational autoencoders (VAE) as anatomical shape generators remain to be explored in terms of utilizing the latent vectors to control or interpret the representativeness of the generated results. In this paper, we propose a pipeline for developing a multi-rib cross-sectional shape generative model from CT images, which consists of the achievement of rib cross-sectional shape data from CT images using an anatomical indexing system and regular grids, and a unified framework to fit shape distributions and associate shapes to anthropometrics for different rib categories. Specifically, we collected CT images including 3193 ribs, surface regular grid is generated for each rib based on anatomical coordinates, the rib cross-sectional shapes are characterized by nodal coordinates and cortical bone thickness. The tensor structure of shape data based on regular grids enable the implementation of CNNs in the conditional variational autoencoder (CVAE). The CVAE is trained against an auxiliary classifier to decouple the low-dimensional representations of the inter- and intra- variations and fit each intra-variation by a Gaussian distribution simultaneously. Random tree regressors are further leveraged to associate each continuous intra-class space with the corresponding anthropometrics of the subjects, i.e., age, height and weight. As a result, with the rib class labels and the latent vectors sampled from Gaussian distributions or predicted from anthropometrics as the inputs, the decoder can generate valid rib cross-sectional shapes of given class labels (male/female, 2nd to 11th ribs) for arbitrary populational percentiles or specific age, height and weight, which paves the road for future biomedical and biomechanical studies considering the diversity of rib shapes across the population.
Collapse
Affiliation(s)
- Yuan Huang
- Research Investigator in International Center for Automotive Medicine (ICAM), University of Michigan, USA.
| | - Sven A Holcombe
- Research Scientist in International Center for Automotive Medicine (ICAM), University of Michigan, USA
| | - Stewart C Wang
- University of Michigan of Surgery and Director of International Center for Automotive Medicine (ICAM), USA
| | - Jisi Tang
- Key Laboratory of Biorheological Science and Technology, Bioengineering College, Chongqing University, China.
| |
Collapse
|
2
|
Holcombe SA, Huang Y, Derstine BA. Population trends in human rib cross-sectional shapes. J Anat 2024; 244:792-802. [PMID: 38200705 PMCID: PMC11021607 DOI: 10.1111/joa.13999] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/25/2023] [Revised: 12/05/2023] [Accepted: 12/12/2023] [Indexed: 01/12/2024] Open
Abstract
Rib fractures remain the most frequent thoracic injury in motor vehicle crashes. Computational human body models (HBMs) can be used to simulate these injuries and design mitigation strategies, but they require adequately detailed geometry to replicate such fractures. Due to a lack of rib cross-sectional shape data availability, most commercial HBMs use highly simplified rib sections extracted from a single individual during original HBM development. This study provides human rib shape data collected from chest CT scans of 240 females and males across the full adult age range. A cortical bone mapping algorithm extracted cross-sectional geometry from scans in terms of local periosteal position with respect to the central rib axis and local cortex thickness. Principal component analysis was used to reduce the dimensionality of these cross-sectional shape data. Linear regression found significant associations between principal component scores and subject demographics (sex, age, height, and weight) at all rib levels, and predicted scores were used to explore the expected rib cross-sectional shapes across a wide range of subject demographics. The resulting detailed rib cross-sectional shapes were quantified in terms of their total cross-sectional area and their cortical bone cross-sectional area. Average-sized female ribs were smaller in total cross-sectional area than average-sized male ribs by between 20% and 36% across the rib cage, with the greatest differences seen in the central portions of rib 6. This trend persisted although to smaller differences of 14%-29% when comparing females and males of equal intermediate weight and stature. Cortical bone cross-sectional areas were up to 18% smaller in females than males of equivalent height and weight but also reached parity in certain regions of the rib cage. Increased age from 25 to 80 years was associated with reductions in cortical bone cross-sectional area (up to 37% in females and 26% in males at mid-rib levels). Total cross-sectional area was also seen to reduce with age in females but to a lesser degree (of up to 17% in mid-rib regions). Similar regions saw marginal increases in total cross-sectional area for male ribs, indicating age affects rib cortex thickness moreso than overall rib cross-sectional size. Increased subject height was associated with increased rib total and cortical bone cross-sectional areas by approximately 25% and 15% increases, respectively, in mid-rib sections for a given 30 cm increase in height, although the magnitudes of these associations varied by sex and rib location. Increased weight was associated with approximately equal changes in both cortical bone and total cross-sectional areas in males. These effects were most prominent (around 25% increases for an addition of 50 kg) toward lower ribs in the rib cage and had only modest effects (less than 12% change) in ribs 2-4. Females saw greater increases with weight in total rib area compared to cortical bone area, of up to 21% at the eighth rib level. Results from this study show the expected shapes of rib cross-sections across the adult rib cage and across a broad range of demographics. This detailed geometry can be used to produce accurate rib models representing widely varying populations.
Collapse
Affiliation(s)
- Sven A. Holcombe
- Morphomics Analysis GroupUniversity of MichiganAnn ArborMichiganUSA
| | - Yuan Huang
- Morphomics Analysis GroupUniversity of MichiganAnn ArborMichiganUSA
| | | |
Collapse
|
3
|
Larsson KJ, Östh J, Iraeus J, Pipkorn B. A First Step Toward a Family of Morphed Human Body Models Enabling Prediction of Population Injury Outcomes. J Biomech Eng 2024; 146:031008. [PMID: 37943113 DOI: 10.1115/1.4064033] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/31/2023] [Accepted: 10/09/2023] [Indexed: 11/10/2023]
Abstract
The injury risk in a vehicle crash can depend on occupant specific factors. Virtual crash testing using finite element human body models (HBMs) to represent occupant variability can enable the development of vehicles with improved safety for all occupants. In this study, it was investigated how many HBMs of different sizes that are needed to represent a population crash outcome through a metamodel. Rib fracture risk was used as an example occupant injury outcome. Morphed HBMs representing variability in sex, height, and weight within defined population ranges were used to calculate population variability in rib fracture risk in a frontal and a side crash. Two regression methods, regularized linear regression with second-order terms and Gaussian process regression (GPR), were used to metamodel rib fracture risk due to occupant variability. By studying metamodel predictive performance as a function of training data, it was found that constructing GPR metamodels using 25 individuals of each sex appears sufficient to model the population rib fracture risk outcome in a general crash scenario. Further, by utilizing the known outcomes in the two crashes, an optimization method selected individuals representative for population outcomes across both crash scenarios. The optimization results showed that 5-7 individuals of each sex were sufficient to create predictive GPR metamodels. The optimization method can be extended for more crashes and vehicles, which can be used to identify a family of HBMs that are generally representative of population injury outcomes in future work.
Collapse
Affiliation(s)
- Karl-Johan Larsson
- Autoliv Research, Vårgårda 447 83, Sweden; Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg 412 96, Sweden
| | - Jonas Östh
- Volvo Cars, Gothenburg 405 31, Sweden; Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg 412 96, Sweden
| | - Johan Iraeus
- Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg 412 96, Sweden
| | - Bengt Pipkorn
- Autoliv Research, Vårgårda SE-44783, Sweden; Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg 412 96, Sweden
| |
Collapse
|
4
|
Robinson A, von Kleeck BW, Gayzik FS. Development and preliminary validation of computationally efficient and detailed 50th percentile female human body models. ACCIDENT; ANALYSIS AND PREVENTION 2023; 190:107182. [PMID: 37390749 DOI: 10.1016/j.aap.2023.107182] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Received: 03/06/2023] [Revised: 05/15/2023] [Accepted: 06/17/2023] [Indexed: 07/02/2023]
Abstract
OBJECTIVE No vehicle testing standard (physical or computational) employs a mid-sized female human surrogate, despite discrepancies related to injury outcomes for female occupants amongst all vehicle users. We detail the design and preliminary validation of 50th percentile female (F50) computational human body models (HBMs) based on Global Human Body Models Consortium (GHBMC) models. METHOD Data for the target geometry was collected as part of the initial generation of GHBMC models. Imaging, surface data, and 15 anthropomorphic measures from a living female subject (60.8 kg and 1.61 m) served as the baseline for model development. Due to the role rib cage geometry plays in biomechanical loading, rib cage morphology from secondary retrospective data was leveraged to identify an average female rib cage based on gross anatomical features. A female rib cage was selected from an existing dataset closest to the mean depth, height, and width of the set, considering only those aged 20 - 50 years. The selected subject among this secondary set also exhibited a 7th rib angle and sternum angle within 5% of the mean measurements, and within the range of previously reported studies. The GHBMC 5th percentile, small female detailed (high biofidelity) and simplified (computationally efficient) models were morphed to match the F50 subject body surface, selected bones, and mean rib cage using established thin plate spline techniques. The models were validated vs. previously published literature studies with an emphasis on rib cage response. Model data was compared to 47 channels of experimental data across four biomechanical hub simulations, two sled test simulations (one of which included all female PMHS), and two robustness simulations to test stability. Model results were mass scaled to the average of the reported corridors. Objective evaluation was conducted using CORA. IRB approval was obtained for all prospective and retrospective data collected or used. The target rib cage was selected from retrospective image data used in prior studies (n = 339 chest CT scans). RESULTS The morphed HBMs closely matched the target geometry. The detailed and simplified models had masses and element counts of 61.2 kg and 61.8 kg, and 2.8 million and 0.3 million, respectively. The mass difference is due to a coarser mesh in the simplified model. The simplified model ran 23 times faster than the detailed model on the same hardware. Each model exhibited stability in robustness tests, and the average CORA scores were 0.80 and 0.72 in the detailed and simplified models, respectively. The models performed well in frontal impacts against PMHS corridors after mass scaling. CONCLUSIONS Numerous recent studies underscore poorer injury outcomes for female vehicle occupants compared to males. While such outcomes are multifactorial, the average female models introduced in this work offer a novel tool within a widely used family of HBMs to reduce the outcome gap in terms of injury for all drivers. HBMs can be deployed in safety studies or in future regulatory requirements faster and more economically than a resized or newly designed ATDs aimed at the same target population.
Collapse
Affiliation(s)
- Andrea Robinson
- Wake Forest University School of Medicine, Department of Biomedical Engineering, United States
| | - B Wade von Kleeck
- Wake Forest University School of Medicine, Department of Biomedical Engineering, United States
| | - F Scott Gayzik
- Wake Forest University School of Medicine, Department of Biomedical Engineering, United States.
| |
Collapse
|
5
|
Holcombe S, Huang Y. Cross-sectional properties of rib geometry from an adult population. Front Bioeng Biotechnol 2023; 11:1158242. [PMID: 37284235 PMCID: PMC10239965 DOI: 10.3389/fbioe.2023.1158242] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/03/2023] [Accepted: 05/08/2023] [Indexed: 06/08/2023] Open
Abstract
Introduction: Human body models (HBMs) play a key role in improving modern vehicle safety systems to protect broad populations. However, their geometry is commonly derived from single individuals chosen to meet global anthropometric targets, thus their internal anatomy may not fully represent the HBM's target demographic. Past studies show sixth rib cross-sectional geometry differences between HBM ribs and population-derived ribs, and corrections to HBM ribs based on these data have improved HBM's abilities to predict rib fracture locations. Methods: We measure and report average and standard deviations (SDs) in rib cross-sectional geometric properties derived from live subject CT scans of 240 adults aged 18-90. Male and female results are given as functions of rib number and rib lengthwise position for ribs 2 through 11. Population means/SDs are reported for measures of rib total area, rib cortical bone area, and rib endosteal area, as well as inertial moment properties of these rib sections. These population corridors are compared between males and females, and against the baseline rib geometries defined in six current HBMs. Results: Total cross-sectional area results found average males ribs to be larger than those of females by between approximately 1-2 SDs depending on rib number and position, and larger in cortical bone cross-sectional area by between 0-1 SDs. Inertial moment ratios showed female ribs being between approximately 0-1 SDs more elongated than male ribs, dependent again on rib number and position. Rib cross-sectional areas from 5 of the 6 HBMs were found to be overly large along substantial portions of most ribs when compared to average population corridors. Similarly, rib aspect ratios in HBMs deviated from average population data by up to 3 SDs in regions towards sternal rib ends. Discussion: Overall, while most HBMs capture overall trends such as reductions in cross-section along shaft lengths, many also exhibit local variation that deviates from population trends. This study's results provide the first reference values for assessing the cross-sectional geometry of human ribs across a wide range of rib levels. Results also further provide clear guidelines to improve rib geometry definitions present in current HBMs in order to better represent their target demographic.
Collapse
|
6
|
Haverfield ZA, Agnew AM, Hunter RL. Differential Cortical Volumetric Bone Mineral Density within the Human Rib. J Clin Densitom 2023; 26:101358. [PMID: 36710221 DOI: 10.1016/j.jocd.2023.01.002] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 07/05/2022] [Revised: 12/18/2022] [Accepted: 01/10/2023] [Indexed: 01/18/2023]
Abstract
INTRODUCTION The human rib provides a vital role in the protection of thoracic contents. Rib fractures are linked to injuries and health complications that can be fatal. Current clinical methods to assess fracture risk and bone quality are insufficient to quantify intra-element differences in bone mineral density (BMD) or to identify at-risk populations. Utilizing quantitative computed tomography (QCT) provides accurate measures of volumetric BMD (vBMD) along the length of the rib which can help delineate factors influencing differential fracture risk. METHODOLOGY One mid-level rib was obtained from 54 post-mortem human subjects (PMHS) and scanned using QCT. Volumes of interest (VOIs) were created for sites at 30%, 50%, and 75% of rib total curve length. Mean Hounsfield units (HU) from each VOI were converted to vBMD using a scan-specific cortical phantom calibration curve. Additionally, rib and lumbar areal BMD (aBMD) were obtained from a sub-sample of 33 PMHS. RESULTS Significant differences in vBMD were found between all sites within the rib (p<0.01). When analyzed by sex, vBMD between the 30% and 50% site were no longer different in either males or females (p>0.05). Separating the sample into discrete age groups demonstrated the relative differences in vBMD between sites diminished with age. Further, age as a continuous variable significantly predicted rib vBMD at all sites (p<0.05), but with little practical or clinical utility (R2, 14.7- 22.8%). Similarly, only small amounts of variation in rib vBMD were explained from DXA lumbar and rib aBMD (R2 , 1.1-21.8%). CONCLUSIONS vBMD significantly decreased from the posterior (30%) site to the anterior (75%) site within the rib which may represent adaptation to localized mechanical loading. These differences could result in differential fracture risk across the rib. As thoracic injury can be fatal, using comprehensive assessments of bone quality that accounts for variation within the rib may provide more accurate identification of at-risk populations.
Collapse
Affiliation(s)
- Zachary A Haverfield
- Injury Biomechanics Research Center, School of Health and Rehabilitation Sciences, The Ohio State University, Columbus, OH, United States.
| | - Amanda M Agnew
- Injury Biomechanics Research Center, School of Health and Rehabilitation Sciences, The Ohio State University, Columbus, OH, United States
| | - Randee L Hunter
- Injury Biomechanics Research Center, School of Health and Rehabilitation Sciences, The Ohio State University, Columbus, OH, United States
| |
Collapse
|
7
|
Molecular age estimation based on posttranslational protein modifications in bone: why the type of bone matters. Int J Legal Med 2023; 137:437-443. [PMID: 36648544 PMCID: PMC9902325 DOI: 10.1007/s00414-023-02948-9] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/21/2022] [Accepted: 01/09/2023] [Indexed: 01/18/2023]
Abstract
Age-at-death estimation is of great relevance for the identification of unknown deceased individuals. In skeletonised corpses, teeth and bones are theoretically available for age estimation, but in many cases, only single bones or even only bone fragments are available for examination. In these cases, conventional morphological methods may not be applicable, and the application of molecular methods may be considered. Protein-based molecular methods based on the D-aspartic acid (D-Asp) or pentosidine (Pen) content have already been successfully applied to bone samples. However, the impact of the analysed type of bone has not yet been systematically investigated, and it is still unclear whether data from samples of one skeletal region (e.g. skull) can also be used for age estimation for samples of other regions (e.g. femur). To address this question, D-Asp and Pen were analysed in bone samples from three skeletal regions (skull, clavicle, and rib), each from the same individual. Differences between the bone types were tested by t-test, and correlation coefficients (ρ) were calculated according to Spearman. In all types of bone, an age-dependent accumulation of D-Asp and Pen was observed. However, both parameters (D-Asp and Pen) exhibited significant differences between bone samples from different anatomical regions. These differences can be explained by differences in structure and metabolism in the examined bone types and have to be addressed in age estimation based on D-Asp and Pen. In future studies, bone type-specific training and test data have to be collected, and bone type-specific models have to be established.
Collapse
|
8
|
Larsson KJ, Iraeus J, Holcombe S, Pipkorn B. Influences of human thorax variability on population rib fracture risk prediction using human body models. Front Bioeng Biotechnol 2023; 11:1154272. [PMID: 37034266 PMCID: PMC10078960 DOI: 10.3389/fbioe.2023.1154272] [Citation(s) in RCA: 6] [Impact Index Per Article: 6.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/30/2023] [Accepted: 03/10/2023] [Indexed: 04/11/2023] Open
Abstract
Rib fractures remain a common injury for vehicle occupants in crashes. The risk of a human sustaining rib fractures from thorax loading is highly variable, potentially due to a variability in individual factors such as material properties and geometry of the ribs and ribcage. Human body models (HBMs) with a detailed ribcage can be used as occupant substitutes to aid in the prediction of rib injury risk at the tissue level in crash analysis. To improve this capability, model parametrization can be used to represent human variability in simulation studies. The aim of this study was to identify the variations in the physical properties of the human thorax that have the most influence on rib fracture risk for the population of vehicle occupants. A total of 15 different geometrical and material factors, sourced from published literature, were varied in a parametrized SAFER HBM. Parametric sensitivity analyses were conducted for two crash configurations, frontal and near-side impacts. The results show that variability in rib cortical bone thickness, rib cortical bone material properties, and rib cross-sectional width had the greatest influence on the risk for an occupant to sustain two or more fractured ribs in both impacts. Therefore, it is recommended that these three parameters be included in rib fracture risk analysis with HBMs for the population of vehicle occupants.
Collapse
Affiliation(s)
- Karl-Johan Larsson
- Autoliv Research, Vårgårda, Sweden
- Division of Vehicle Safety, Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg, Sweden
- *Correspondence: Karl-Johan Larsson,
| | - Johan Iraeus
- Division of Vehicle Safety, Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg, Sweden
| | - Sven Holcombe
- International Center for Automotive Medicine, University of Michigan, Ann Arbor, MI, United States
| | - Bengt Pipkorn
- Autoliv Research, Vårgårda, Sweden
- Division of Vehicle Safety, Department of Mechanics and Maritime Sciences, Chalmers University of Technology, Gothenburg, Sweden
| |
Collapse
|