Upload
hoangduong
View
221
Download
2
Embed Size (px)
Citation preview
Abstract Costal cartilage bridges the sternum and the ribs and plays a key role in the biomechanics of the
chest. Costal cartilage is known to calcify in local regions with age, which can substantially stiffen its overall
response to loading. However, the rate of accumulation of the calcified volume is not well quantified. Current
computational models of the thorax assign a homogeneous soft material to cartilage segments, yet their finite
element meshes are well suited to the specification of interstitial calcification zones, should applicable data
become available.
This study measures volumes and extents of costal cartilage calcification from 205 live subject CT scans.
Significant increases in volume calcification – both in a given cartilage segment and in the lengthwise extent of
those segments that experience calcification – are seen with age (p<0.0001). Age and sex accounted for 35% of
all inter‐individual population variability. Specific recommendations for introducing person‐age via regional
calcification to models of the costal cartilage are that (1) calcification volume within a segment should increase
at the rate of 0.9 mm (for an equivalent cube edge length) per decade, and (2) should involve an increasing
lengthwise extent of the cartilage segment at a rate of at least 7% per decade.
Keywords age, calcification, cartilage, costal cartilage calcification, rib.
I. INTRODUCTION
The chest plays a major role in mitigating the forces that the body experiences during motor vehicle crash
events. Costal cartilage is a key structural component of the chest, contributing to its rigidity and reinforcing the
chest wall by bridging the space between the sternum and the ribs. Simulated stiffening of the costal cartilage
through an added tough outer shell has been shown to increase the overall chest stiffness by 50% [1], and
simulated stiffening of the cartilage material itself has increased subsequent stress levels in the ribs by 63%
during loading of the sternum [2] and increased rib fractures in diagonal belt loading [3].
Costal cartilage is organised in a series of segments connected to the ribs, with typical topology connecting
ribs 1‐4 directly to the sternum, ribs 5‐7 to both the sternum and to superior segments, and ribs 8‐10 to only
superiorly neighbouring segments. Its material structure includes a hyaline cartilage mid‐substance surrounded
by an outer layer of perichondrium. Over time, portions of the costal cartilage can calcify (less commonly but
more precisely termed ossify), producing inhomogeneities of substantially harder and stiffer material properties
within or around a cartilage segment. This calcification can substantially stiffen the mechanical response of the
cartilage, particularly when the calcified tissue forms a bridge across the length of a costal segment [4].
An association between advancing age and presence (or size) of calcification is well documented [5‐9], and
studies have also described differences between the typical patterns of calcification seen between males and
females [8,10]. Results from these studies have been used forensically to help estimate the age and sex of
unknown deceased persons [8,11]. However, with the most common data source being X‐ray images taken from
an anterior aspect, there is limited volumetric information describing the typical amounts and distribution of
calcification that might inform 3D biomechanical models.
Computational human body models are increasingly being used to understand the biomechanics of the
thorax to loading, and recent efforts have included the concept of person‐age by building models specifically
representing elderly individuals [3,12‐14]. Forman and Kent [15] have highlighted the need for costal cartilage
calcification to be considered when modelling changes in the mechanics of the rib cage with age, and have
explored a methodology for the placement of calcified elements within current model meshes. One aspect still
SA. Holcombe, PhD ([email protected], +17346475444) and S. Ejima, PhD are Research Scientists at the University of Michigan in the International Center for Automotive Medicine (ICAM), USA. SC. Wang, PhD, MD, is a University of Michigan Professor of Surgery and director of ICAM, USA.
Calcification of Costal Cartilage in the Adult Rib Cage.
SA. Holcombe, S. Ejima, SC. Wang
IRC-17-106 IRCOBI conference 2017
-737-
lacking, however, is an understanding of how much of a given cartilage segment should be specified as calcified
within a model in order to best represent the costal cartilage of a given demographic.
This study aims to contribute to the body of knowledge required to address that need by quantifying the
volume and extent of cartilage calcification that is seen in adults, and to describe the population variability that
exists for such measures. Specifically, this study aims to quantify the rate of calcification accumulation with age
that is typical within a live adult population, and to quantify the differences in calcification amount that is
expected between males and females. Such information can help build more accurate models of calcified
cartilage in order to better understand the effect that age has on the biomechanics of the chest.
II. METHODS
Scan and Study Population
The study population consisted of 205 individuals (110 female, 95 male) aged 17 to 91 years. All subjects had
received chest CT scans obtained for trauma purposes between 2001 and 2016, collected from the International
Center for Automotive Medicine (ICAM) morphomics database. All scans were taken in a supine position. Scan
axial resolution ranged from 0.56 mm to 0.97 mm, and scan slice spacing ranged from 0.625 mm (53% of scans)
to 5.0 mm (2% of scans) with 93% of scans taken at 1.25 mm spacing or lower. 47 subjects with rib fractures
were retained in the study, however subjects with extreme skeletal abnormalities or fixation devices present in
the scan were excluded.
Costal Cartilage Identification
To identify regions of costal cartilage calcification from within the CT image volumes a set of curved 3D
reformations through those image spaces were developed via custom‐made routines in MATLAB 2016b (The
Mathworks Inc., Natick, MA). These reformations serve to spatially isolate the anterior chest wall (containing
the costal cartilage segments and the sternum) from other parts of the body. Within this space, traditional
histogram‐based segmentation and morphological operations can then extract the volumetric regions of locally
calcified cartilage.
The reformations are based on a grid of landmarks that are placed manually into the 3D space of the CT scan
at (1) the ends of left and right ribs (designating the start of each costal segment), (2) the sternal connection
locations (or ends of each costal segment), and (3) a point approximately mid‐way along each costal segment at
its cross‐sectional centre. A curved 3D surface is then generated that passes smoothly through this grid of
landmarks [16], forming an approximate representation of the mid‐surface of the anterior chest wall as depicted
in Fig. 1 (left). A volumetric region encompassing the costal cartilage is identified as the 3D space within a 10
mm offset from this mid‐surface. Anatomically, costal cartilage segments meet the ribs at a cup‐shaped
depression in the rib cortical bone, and the rib end landmark is placed specifically at the apex of this cup [4].
Similarly, a depression is present at the sternal end of the cartilage segments and the apex of each cup is chosen
for the placement of each sternal connection landmark. To ensure accuracy, custom MATLAB interfaces were
written allowing a user to place an approximate landmark on regular axial CT slices, then adjust that landmark
using multiply resliced oblique views centred on that local region.
Image voxels for the 2nd through 5th rib cartilage segments were resampled through this curved volumetric
region from the original CT data using bilinear interpolation. The zone between each sternal landmark was
excluded, and the remaining space was partitioned into individual segments for each rib level based on
proximity to the central curve from each rib end to sternal connection point. A threshold of 180 Hounsfield
Units (HU) was used to separate calcified regions of cartilage from the other (lower density) surrounding tissues.
Minor morphological operations then removed extremely small calcification regions (less than 1 mm3) and to
remove the internal thoracic arteries which run vertically adjacent to the sternum, producing the isolated
regions of calcification as illustrated in Fig. 1 (right). Note that under this overall segmentation scheme the small
amount of bone contiguous with the rib or the sternum which forms the connecting cup (which is on the order
of 5 mm deep) will also be classified as calcification.
IRC-17-106 IRCOBI conference 2017
-738-
Fig. 1. Example mid‐surface of the chest wall through cartilage segments built from the grid of manually placed rib end, sternum connection, and mid‐segment landmarks (left). In this view costal cartilage calcification is clearly visible as white regions along segments. Full volumetric regions from the same subject are shown in 3D (right).
Measured Variables and Statistical Analyses
The primary object of measurement was the volumetric amount of calcification identified within each
cartilage segment space, and this volume, V, was recorded for each segment in units of mm3. To improve data
normality for statistical analyses, and to retain conceptual understanding, volumetric results are presented in
this study by a cubed root transformation to the edge length E (mm) of a cube containing the equivalent volume
of calcification as was measured from within the costal segment. A secondary measurement for the extent of
calcification along a segment was characterised by the lengthwise percentage (LenPcnt, %) of each costal
segment that contained any calcified region within its cross section. For example, a segment with calcification
regions emanating from the ribs and from the sternum each involving some portion of 20 % of the segment’s
overall length would be designated with a lengthwise percentage (LenPcnt) of 40 %.
These measurements were taken from eight separate cartilage segments (ribs 2 through 5, left and right
sides of the body), amounting to a total of 1640 observations. Results are reported firstly on a per‐segment
basis and, for statistical analyses, the average E and LenPcnt values are taken across all segments per individual
reducing the data to one observation per individual. Multivariable linear regression is performed to investigate
associations between measured variables and subject age and sex, with statistical significance reported at the
p<0.05 level.
III. RESULTS
Measurements of calcification volume were collected from all costal cartilage segments. Fig. 2 shows the
population distribution of equivalent cube edge length (E, mm) of these volumes for males and females,
stratified by rib segment level. These distributions show that males typically have a higher calcification volume
than females at each segment level (p<0.01). Lower rib levels (4 and 5) are also seen to contain a marginally
greater volume of calcification than levels 2 and 3 (with E increased by 2.1 mm, p<0.01). For reference, the
average segment lengths are given in Fig. 2, showing a clear trend for lower cartilage segments being longer
than upper segments at increments of approximately 8 mm. When considering the percentage length of rib
segments that show calcification involvement (LenPcnt), there are no longer clear sex differences or differences
in extent calcification by rib segment level.
IRC-17-106 IRCOBI conference 2017
-739-
Fig. 2. Results per segment for the overall calcification volume (given by the equivalent cube edge length, E), the geometric length of each segment (mm), and the percentage length of a cartilage segment with non‐zero calcification involvement (LenPcnt). Boxes show the median and the 25th to 75th quantile ranges and whiskers extend to the ±2.7 sigma range (99.7% coverage).
The measurements of calcification volume (E) were averaged across the eight costal segments for each
individual, and are shown by age and sex in Fig. 3 below. Linear regression trend lines are also shown, with
regression coefficients summarised in Table 1. Regression results showed significant increases (p<0.0001) in the
volume calcification (with equivalent cube edge length, E, increasing at the rate of 0.9 mm per decade) and
extent calcification with age (additional 6.9 % of segment length involvement per decade). Males had higher
overall volumes of calcification (average E across all segments being 1.56 mm larger, p<0.0001) than females,
however no significant difference in extent calcification () was seen between sexes (p>0.1). Results also did not
support any sex‐based differences in the rate of calcification accumulation with age, as subsequent regression
models with an additional age‐sex interaction term found that term to not be significantly different to zero
(p>0.1).
IRC-17-106 IRCOBI conference 2017
-740-
Fig. 3. Average calcification volume (equivalent cube edge length, E) per segment by age (top left), and average lengthwise extent of calcified cartilage (LenPcnt, top right). Chest wall mid‐surface images for individuals marked A‐C are shown for comparison (bottom row).
TABLE I MULTIVARIABLE REGRESSION RESULTS TO CALCIFICATION MEASUREMENTS
E (mm) r2 = 0.35 LenPcnt (%) r2 = 0.26
Predictor Estimate p Estimate p
Intercept 2.545 < 0.0001 21.805 < 0.0001
Sex (0=M,1=F) ‐1.567 < 0.0001 ‐5.502 0.13
Age (years) 0.089 < 0.0001 0.689 < 0.0001
IV. DISCUSSION
This study reports quantitative measurements of costal cartilage calcification taken from CT images from a
population of live adult subjects. There has been previous in‐depth quantitative analysis of cartilage segments
from 11 PMHSs at the microstructure level using MicroCT [4], and limited quantitative data from analysis of
large populations for the purpose of age and sex determination using X‐ray imaging [9,10]. The measurements
presented here are intended to assist researchers design calcification schemes that can be added to current
human body computational models in order to better model the aging process within the costal cartilage.
In part due to a limited amount of available information, current thoracic computational models assign a
single homogeneous material to the costal cartilage, with material properties representative of the hyaline
cartilage mid‐substance without calcification [17‐19]. Such finite element models, however, have the capacity to
assign specific material properties to individual finite elements within a solidly meshed part. Furthermore, shell
elements can be added to cartilage surface parts to form a separate perichondrium layer, again capable of non‐
uniform material properties.
Forman and Kent [20] implemented an inhomogeneous cartilage material into subject‐specific finite element
component models of the costal cartilage. Their modelling schematic included setting the distinct solid finite
elements corresponding to regions of calcification (as derived from CT images of the cadaveric specimens being
modelled) to a stiffer material property. Forman and Kent [15] further expanded this methodology by
introducing a parametric calcification severity, whereby progressively larger (or smaller) clusters of elements
were assigned the calcification material model compared to a baseline individual. The calcification scoring
system used was based partly on methods developed by McCormick [7] from anterior X‐ray images. The
A B C
IRC-17-106 IRCOBI conference 2017
-741-
technique used here instead utilises 3D volumetric data from CT images to quantify calcification amounts. This
could advance the modelling methodology (whereby specific finite elements are set to a calcified material
property) by setting volumetric targets explicating the amount that a calcification region should expand in order
to provide a target volume of calcification and consequently represent a chosen increase in age.
Microstructural studies have shown that calcification can occur in multiple different structured morphologies
including, in some cases, dense shells surrounding a core of struts similar to trabeculae [4]. While not possible to
discern under clinical CT scan resolutions, these fine details are also not easily represented in current full body
computational models (which have cartilage element edge lengths in the order of 2‐10 mm). This study has
consequently focused on quantifying the overall volumes of calcification within a segment and understanding a
simplified representation of its spatial distribution. Analytic work has explored microstructural connectivity
variations in cartilage, simulating the elastic modulus of heterogeneous cartilage segments with calcified regions
assigned either in clusters or dispersed throughout a cartilage segment [21]. Results showed that stiffness
increases substantially as calcification connectivity increases, suggesting that FE models would exhibit different
behaviour based both on the amount of calcification imposed, and on its spatial configuration within a cartilage
segment.
The volumes of calcification reported in the current results are consistent with expectations based on
previous literature. Most studies report that calcification typically begins to form in approximately the third
decade of life [9,22], is commonly of moderate severity between ages 50–60 [5,7], and is moderate or severe in
approximately 45% of persons over 75 [7,23]. The current data largely reflects these trends, however it is
notable that the lowest end of the calcification volume distribution does not actually reach zero. This is likely
due to the relatively simplistic delineation between rib bone and costal cartilage that is used here, whereby the
cartilage and rib or sternal bones are separated via an assumption about their connections’ cup‐shaped
morphology. This assumption means that the periphery of this cup‐shaped depression likely contains small
volumes of rib bone which are being considered as calcified volume, making the overall calcification volume
non‐zero.
Observation of the calcification patterns seen from mid‐surface reformations of the chest wall (see Fig. 3
second row, for example) match well with previous descriptions of sex‐based differences in the morphology of
accumulated calcified cartilage. Specifically, males are reported to exhibit a peripheral pattern characterised by
ossification of the inferior and superior costal cartilage margin, whereas females show one or more central
calcification patterns with pyramidal‐shaped central tongues of ossification or centrally‐placed smooth globules
of ossification [6,22]. Such differences are seen clearly in the two highly calcified individuals highlighted in Fig. 3
(lower row), and similar patterns were observed across most other individuals. Cartilage calcification occurs as
calcium uptake into the cartilage tissue increases. The underlying mechanism for this difference between males
and females in costal cartilage calcification patterns has not been fully explored. However, with calcium uptake
into cartilage tissue being the primary cause of calcification in general, early hypotheses have come from finding
similar dimorphisms in cartilage presentation near the thyroid gland suggesting hormonal differences between
males and females may be involved [6,24,25], as well finding high calcium concentrations in sections of the
aorta more frequently in males than females [25].
In terms of its application to computational models, one limitation of the data presented here comes from
the fact that it is based on the segmentation of calcified regions of the costal cartilage, but without an
accompanying segmentation of the overall cartilage segment. The reason for this is that the difference in radio‐
density between costal cartilage and its various surrounding tissues (such as muscle, fat) is relatively small, such
that a simple histogram‐based threshold is unable to clearly separate these body components. Cartilage
calcification, on the other hand, is considerably denser than other local materials and as such it can be easily
segmented after localising the containing region for a given segment of costal cartilage. The consequence of
measuring only the calcified regions within a segment is that no true volumetric percentage of calcification can
be reported. As a surrogate for this more desirable measurement, we have utilised the costal segment length
(which does not require segmentation of a full cartilage segment) and chosen to report a lengthwise percentage
of calcification as a guide for researchers choosing a scheme for calcification growth along the cartilage
segment.
IRC-17-106 IRCOBI conference 2017
-742-
V. CONCLUSIONS
Measurements of costal cartilage calcification volumes and distribution have shown significant trends with age. Age and sex are seen to account for around 35% of all inter‐individual population variability in the total volume of calcified cartilage. The lengthwise extent of calcified regions of a cartilage segments is also seen to increase significantly with age. Results provide volumetric guidelines for the introduction of regional calcification within the costal cartilage. Specifically, it is recommended when incorporating person‐age into models that the calcified cartilage volume should increase at the rate of 0.9 mm (for an equivalent cube edge length) per decade, and should involve an increasing lengthwise extent of the cartilage segment at a rate of at least 7% per decade.
VI. REFERENCES
[1] Murakami, D., Kobayashi, S., Torigaki, T., and Kent, R. Finite element analysis of hard and soft tissue contributions to thoracic response: sensitivity analysis of fluctuations in boundary conditions. Stapp Car Crash Journal, 2006. 50: p. 169‐189
[2] Oyen, M.L., Murakami, D., and Kent, R. Mechanical characterization of costal cartilage. Proceedings of 33rd International Workshop on Human Subjects for Biomechanical Research, 2005.
[3] Antona‐Makoshi, J., Yamamoto, Y., et al. Age‐Dependent Factors Affecting Thoracic Response: A Finite Element Study Focused on Japanese Elderly Occupants. Traffic Injury Prevention, 2015. 16(sup1): p. S66‐S74
[4] Lau, A., Kindig, M., and Kent, R. Morphology, distribution, mineral density and volume fraction of human calcified costal cartilage. Acta Biomaterialia, 2011. 7(3): p. 1202‐1209
[5] Dearden, L.C., Bonucci, E., and Cuicchio, M. An investigation of ageing in human costal cartilage. Cell and Tissue Research, 1974. 152(3): p. 305‐337
[6] Elkeles, A. Sex differences in the calcification of the costal cartilages. Journal of the American Geriatrics Society, 1966. 14(5): p. 456‐462
[7] McCormick, W.F. Mineralization of the costal cartilages as an indicator of age: preliminary observations. Journal of forensic sciences, 1980. 25(4): p. 736‐741
[8] Stewart, J.H. and McCormick, W.F. A sex‐ and age‐limited ossification pattern in human costal cartilages. Am J Clin Pathol, 1984. 81(6): p. 765‐9
[9] Teale, C., Romaniuk, C., and Mulley, G. Calcification on chest radiographs: the association with age. Age and ageing, 1989. 18(5): p. 333‐336
[10] Navani, S., Shah, J., and Levy, P. Determination Of Sex By Costal Cartilage Calcification. American Journal of Roentgenology, 1970. 108(4): p. 771‐774
[11] Ontell, F.K., Moore, E.H., Shepard, J.A., and Shelton, D.K. The costal cartilages in health and disease. Radiographics, 1997. 17(3): p. 571‐7
[12] Kimpara, H., Iwamoto, M., et al. Effect of Assumed Stiffness and Mass Density on the Impact Response of the Human Chest Using a Three‐Dimensional FE Model of the Human Body. Journal of Biomechanical Engineering, 2006. 128(5): p. 772
[13] El‐Jawahri, R., Laituri, T., Ruan, J., Rouhana, S., and Barbat, S. Development and validation of age‐dependent FE human models of a mid‐sized male thorax. Stapp car crash journal, 2010. 54: p. 407‐430
[14] Hwang, E., Hu, J., et al. Development, Evaluation, and Sensitivity Analysis of Parametric Finite Element Whole‐Body Human Models in Side Impacts. Stapp car crash journal, 2016. 60: p. 473‐508
[15] Forman, J. and Kent, R. The effect of calcification on the structural mechanics of the costal cartilage. Computer Methods in Biomechanics and Biomedical Engineering, 2014. 17(2): p. 94‐107
[16] D’Errico, J. Inpaint NaNs. 2004: MATLAB Central File Exchange, URL: https://www.mathworks.com/matlabcentral/fileexchange/4551. Accessed 2017 Feb 20.
[17] Furusua, K., Watanabea, I., Katoa, C., Mikia, K., and Hasegawab, J. Fundamental study of side impact analysis using the finite element model of the human thorax. JSAE Review, 2001. 22(2): p. 195‐199
[18] Behr, M., Arnoux, P.J., et al. A Human Model for Road Safety: From Geometrical Acquisition to Model Validation with Radioss. Computer Methods in Biomechanics and Biomedical Engineering, 2003. 6(4): p. 263‐273
[19] Ruan, J., El‐Jawahri, R., Barbat, S., and Prasad, P. Biomechanical Analysis of Human Abdominal Impact Responses and Injuries through Finite Element Simulations of a Full Human Body Model. Stapp Car Crash Journal, 2005. 49: p. 343‐366
IRC-17-106 IRCOBI conference 2017
-743-
[20] Forman, J. and Kent, R. Modeling costal cartilage using local material properties with consideration for gross heterogeneities. Journal of Biomechanics, 2011. 44(5): p. 910‐916
[21] Lau, A., Kindig, M., Salzar, R., and Kent, R. Micromechanical modeling of calcifying human costal cartilage using the generalized method of cells. Acta Biomaterialia, 2015. 18: p. 226‐235
[22] Rejtarova, O., Slizova, D., Smoranc, P., Rejtar, P., and Bukac, J. Costal cartilages ‐ a clue for determination of sex. Biomedical Papers, 2004. 148(2): p. 241‐243
[23] Edge, J.R., Millard, F.J.C., Reid, L., and Simon, G. The Radiographic Appearances of the Chest in Persons of Advanced Age. The British Journal of Radiology, 1964. 37(442): p. 769‐774
[24] Hatley, W., Samuel, E., and Evison, G. The pattern of ossification in the laryngeal cartilages: A radiological study. The British journal of radiology, 1965. 38: p. 585‐591
[25] Kanabrocki, E.L., Fels, I.G., and Kaplan, E. Calcium, cholesterol, and collagen levels in human aortas. J Gerontol, 1960. 15: p. 383‐7
IRC-17-106 IRCOBI conference 2017
-744-