Temporal mapping of the closure of the anterior fontanelle and contiguous sutures using computed tomography, in silico models of modern infants

Abstract

The aim of this study is to quantify and statistically model the age‐related decline in the fibrous connective tissue interface of the anterior fontanelle in modern Australian infants, using three‐dimensional, semi‐automated computed‐assisted design protocols. Non‐linear regression with variance models, using power functions, combined with quantile regression of the 5th and 95th population percentiles, were utilised to assess absolute anterior fontanelle surface area (AFSA) as a function of age, using multi‐slice cranial computed tomography scans obtained from 256 infants aged < 30 months (males: n = 126, females: n = 109) from Brisbane children’s hospitals. Normalised AFSA (NFSA), standardised for variation in cephalic size, followed a progressive decline from birth, the greatest velocity change occurring between the 3–6 and 6–9 month cohorts. Growth of the neurocranium is the most significant within the first 8 months postpartum, with a mean increase of 19.03 mm in maximum cranial length and 10.04 mm in breadth. Directionality of fontanelle closure, quantified using spline curves refutes fundamental assumptions that the anterior fontanelle is consistent with a quadrilateral, and contiguous sutures exhibit constant velocity of closure. The present study provides normative values for fontanelle size and diameters as well as new predictive non‐linear models for age substantiation, screening of developmental abnormalities and indicators of suspected child maltreatment in modern infants aged birth to 30 months.

This study provides the first series of normative data for absolute and normalised surface area, as well as diameters for the anterior fontanelle in modern Australian multi‐ancestral infants aged 2 years and under. We provide predictive functions to assess paediatric aberrations in suspected child maltreatment, living age substantiation and assessment of craniofacial growth and development. These new standards reduce temporality and demographic biases inherent in current methods, and the quantification of suture closure refutes the literature on directionality and morphology.

1. INTRODUCTION

The neonate cranium features a number of connective tissue fontanelles that juxtapose the primary ossification centres of cranial bones (Figure 1). The anterior fontanelle (AF) represents the residual membranous remnants of the ectomeninx, the neural crest cell‐derived tissue from which the calvarial bones develop (Jiang et al., 2002). The dimensions of the AF serve as an index of the rate of development and ossification of the calvarium and may be indicative of altered intracranial pressure (Dechant et al., 1999; Mathijssen et al., 1999). Several factors contribute to the variability in morphology of the AF, including gestational age (Davies et al., 1975; Adeyemo and Omotade, 1999), sex (Popich and Smith, 1972; Srugo and Berger, 1987; Mir and Weislaw, 1988) and ancestry (Dubowitz et al., 1970; Davies et al., 1975; Philip, 1978; Faix, 1982; Ogunye et al., 1982; Mir and Weislaw, 1988; Chang and Hung, 1990; Lyall et al., 1991; Mattur et al., 1994; Omotade et al., 1995). Geographical variation in fontanelle size within Nigerian (Ogunye et al., 1982) and Indian (Chakrabarti, 1989; Mattur et al., 1994) infants has been reported, validating the need for population‐specific reference data. Skeletal maturation standards should therefore not be extrapolated from one population to another, with secular change in accelerated maturation rates (Malina, 2004; Heuze and Cardoso, 2008; Langley‐Shirley and Jantz, 2010) underscoring the importance of also establishing contemporary standards for the examination of modern infants. Lottering et al. (2016) provide recalibrated ossification timings of the primary centres of the calvarium and select cervical vertebrae in modern Australian subadults, reporting complete closure of the AF between 14 and 19 months in males, and 16 and 22.5 months in females, at a 95% credible interval using Bayesian modelling. Concurrently, the width of the metopic suture starts decreasing (< 2 mm) after 6 days post‐birth in males and 12 days in females, prior to closing between 1 and 2 months, respectively. In an investigation of ethnic differences in time and rate of closure, Kirkpatrick et al. (2019) report that the AF is closed by 19 months in New Zealand (NZ) Maori/Pacific infants (n = 116), whereas 50% of NZ European infants (n = 47) still present with remnants of an AF in the 19–24 month cohort. Considerable variation in fontanelle size is evident within and across age cohorts, possibly attributable to a small sample size.

FIGURE 1.

Anatomical representation of the osteology and connective tissue network in a neonatal skull (newborn), in anterior (left), superior (middle) and posterior (right) views. AF, anterior fontanelle; PF, posterior fontanelle; MS, metopic suture; SS, sagittal suture; LCS, left coronal suture; RCS, right coronal suture; LLS, left lambdoidal suture; RLS, right lambdoidal suture

The Handbook of Physical Measurements (Hall et al., 2006) commonly referenced in modern clinical examination, provides the following definition for AF landmark identification and measurement, based on the original work by Popich and Smith (1972):

The index finger should be placed as far as possible into each of the four corners of the anterior fontanelle. These four positions may be marked with a dot immediately distal to the fingertip.... The points are joined to form a quadrilateral, and the sum of the longitudinal and transverse diameters along the sagittal and coronal sutures can be measured.

Hall et al. (2006) note the difficulty in defining the boundary landmarks using the aforementioned definition, due to the wide patency of the sagittal and metopic sutures which communicate with the adjacent metopic and posterior fontanelles within the first months of life. Furthermore, delineation of the AF through palpation and determination of absolute size may be confounded by variation in the width of the practitioner’s fingertips. Uzukwu‐Edeani et al. (2013) propose that reported AF size differences between a number of regional sub‐populations in Nigeria (Ogunye et al., 1982; Adeyemo et al., 1999) may be attributed to methodological inconsistencies between studies rather than actual population differences, emphasising the need for improved AF measurement practices that are objective and accurate. The aforementioned inconsistencies pertain to the delineation of the AF boundaries, which are not clearly stated in the Adeyemo et al. (1999) and Ogunye et al. (1982) studies, and the measurement strategy employed, i.e. whether infants aged ≤ 48 hr with over‐riding cranial bones (from excessive parturition moulding) were included/excluded. An alternate method of quantifying AF size through calculating the surface area of the fontanelle in square millimetres using trigonometry of a triangle has been proposed by Davies et al. (1975). Both linear and areal dimensions rely on the fundamental assumption that the fontanelle morphology is consistent with the shape of a quadrilateral, which is an over‐simplification of its three‐dimensional curvilinear structure and is based on the assumption that the contiguous sutures close at the same rate.

The assessment of AF surface area (AFSA) has become common in fetal and post‐natal examination, due to the advancement and application of three‐dimensional ultrasonography for in utero assessment of craniofacial morphology (Chaoui et al., 2005; Faro et al., 2005, 2006). Irregular development of the AF may be indicative of abnormal craniofacial growth (Davies et al., 1975; Philip, 1978; Paladini et al., 2007) and the diagnosis of systemic disorders. For example, reduced AFSA is common in fetuses with craniosynostosis of the sagittal and/or coronal sutures (Kreiborg et al., 1993; Cohen and MacLean, 2000); whereas large AFSA is associated with chromosomal disorders (e.g. Down syndrome, Paladini et al., 2007; 2008), congenital hypothyroidism (Smith and Popich, 1972) and skeletal disorders such as achondroplasia and osteogenesis imperfecta (Smith and Popich, 1972). Irregular AFSA may also be attributed to cardiac and central nervous system malformations (Paladini et al., 2008). Within Australian paediatric assessment, quantification of the AF is currently not included in the standard paediatric toolkit in Queensland public hospitals. Although subjective and highly dependent on practitioner experience, palpation is the preferred examination technique, with ‘bulging’ or ‘sunken’ fontanelle status of primary interest. This presumptive approach may be considered more time‐ and/or cost‐efficient and avoids unnecessary referrals for medical imaging; if preliminary screening suggests cause for concern, further examination may then be justified.

In an attempt to standardise examination of the cranial fontanelles, the current study utilises computed tomography imaging to quantify this connective tissue zone using a semi‐automated methodological approach developed by Lottering et al. (2014) to assess the maturation of the AF and adjoining sutures (metopic, coronal and sagittal) with respect to size, closure and sexual dimorphism from birth to 30 months postpartum in trauma‐screened Australian infants. The aim of this study is to provide baseline measurements of the surface area and maximum linear dimensions of the AF for evaluating growth and development, in the form of predictive statistical models for paediatric screening of developmental abnormalities, disease states and living age estimation of individuals lacking identification.

2. MATERIALS AND METHODS

2.1. Sample demographics and data acquisition

The sample consisted of de‐identified retrospective cranial/cervical spine DICOM datasets acquired from 235 multi‐ancestral individuals (males: 126, females: 109) aged neonate to 30 months postpartum, subject to multi‐slice computed tomography (MSCT) scanning in the Departments of Medical Imaging at Queensland (Australian) Hospitals from 2010 to 2014. Thin‐slice DICOM data from children scanned under the ‘Trauma Activation Criteria’ were sourced from the Enterprise PACS database, onsite at the Lady Cilento Children’s Hospital (formerly Royal Children’s Hospital) and Mater Health Services (Mater Adult Hospital and Mater Children’s Hospital), prior to anonymisation in OsiriX® (Version 4.1–64 bit) (Visage Imaging GmbH). Scans were conducted at the Royal Children’s Hospital using a Toshiba® Aquilion LBTM 64‐slice machine (Toshiba Medical Systems; voxel size: 0.391 × 0.391 × 0.30) and at Mater Health Services using a Philips® Brilliance 64‐slice machine (Philips Medical Systems; voxel size: 0.363 × 0.363 × 0.45; 135 kV, mA varied according to ‘Sure Exposure’). A demographic breakdown of the Enterprise PACS database, providing the percentage of scans acquired from each affiliated hospital and respective geographical location, is disseminated in Lottering et al. (2017). Biological parameters including the patient’s date of birth, date of scan, sex, weight and postcode of residence were embedded into the meta‐data of each dataset and disclosed to investigators following blind analysis. Ethical approval was granted by The Children’s Health Service District Ethics Committee (HREC/12/QRCH/208), ratified by The Queensland University of Technology Research Ethics Unit (Approval No. 1,200,000,716) and approved by The Queensland Government under the Public Health Act (Section 284) 2005 (RD004657).

Exclusion criteria. Neonates ≤ 38 weeks’ gestation, infants presenting with documented craniosynostosis, bulging fontanelles, hydrocephalus or excessive moulding of the cranial bones, as well as patients presenting with orthopaedic materials (i.e. screws, shunts), foreign bodies in the scan field and displaced complex fractures across the sutures were excluded. The ancestry of each patient is not recorded upon clinical examination, as it is deemed medically irrelevant, with the exception of individuals who self‐identify as Australian Aboriginal or Torres Strait Islander, which is collected for health survey purposes. We make the assumption that our sample reflects the ancestral composition detailed in the Queensland Census results (Australian Bureau of Statistics, 2011), reporting that 56.8% of primary responses ascertain European ancestral origins (within two generations), followed by 26.3% of Australian (sub‐classification of Oceanian ancestry) and 5.5% of Queensland individuals reported being of Asian ancestry. Individuals identifying as Australian Aboriginal or Torres Strait Islander were excluded due to minority status (constituting only 6.1% of children aged 0–4 years and 5.8% aged 5–14 years in Queensland) and under‐representation on the Enterprise PACS server.

2.2. Data transformation

Thin‐slice stacked DICOM sets were uploaded into Amira® (VSG, FEI Company) for image segmentation and isosurface model generation. Isosurface models for each infant were extracted as a stereolithography (stl) file native to computer‐aided design (CAD) software including geomagic design X^TM (3D Systems, Inc.,). In geomagic design X^TM, each stl model was scaled and cleaned by interpolating missing surface triangles/voxels and applying a smoothing algorithm to correct any error introduced in the isosurfacing process, prior to optimisation to produce a high‐quality three‐dimensional (3D) reconstruction (Lottering et al., 2014). To determine absolute dimensions, including maximum length (GOL), breadth (XCB) and basio‐bregmatic height (BBH), the morphometric protocol developed by Lottering et al. (2014) utilising extreme position planes of a sagittal silhouette curve was implemented. The non‐delineated maximum sagittal/longitudinal (SLD) and coronal/transverse (CLD) diameters of the fontanelle were also calculated using geomagic design X^TM measurement capabilities. In the case of open sutures, these measurements extended into the contiguous sutures. Further, as per the methodology of Lottering et al. (2014), the absolute fontanelle area (AFSA) constituting the connective tissue interface of the anterior fontanelle and adjoining sagittal, coronal and metopic sutures was quantified in mm². To account for relative size differences of the cranial vault between individual samples, each area was normalised as a percentage of the superior surface model dimensions GOL and XCB to calculate the normalised fontanelle surface area (NFSA). The following trigonometry formula for an ellipse was applied for this transformation:

$$\text{NFSA}=\frac{\text{AFSA}}{\pi \left(r_1\right)\left(r_2\right)}\times 100.$$

where AFSA = absolute surface area of the anterior fontanelle and contiguous sutures, in mm; r ₁ = GOL/2 and r₂ = XCB/2. Additional quantification of the surrounding sutures and fontanelles allowed us to assess the sequence and directionality of closure of the connective tissue network. Utilising the 3D mesh sketch function, 3D spline measurements of the total length of each suture were conducted, accompanied by the length constituting unossified fibrous tissue. These measurements were transformed into the percentage of the suture that had closed.

2.3. Statistical analysis

Validation of the protocols employed in this study, quantified using measurement error variability (technical error of measurement [TEM] and relative TEM) and retest reliability (intra‐class correlation coefficient) statistics, are presented in Lottering et al. (2014) and demonstrate high levels of inter‐rater reliability, exceeding recommended anthropological standards (Ulijaszek and Kerr, 1999). Descriptive statistics, including the mean and standard deviation for each age cohort, were calculated for each sex. Due to a non‐parametric distribution across all variables, sex and age effects were calculated using Mann–Whitney U‐tests (p < .05 significance level).

Non‐linear regression was used to model the data in R (R Development Core Team, 2013) using the nlme library (Pinheiro et al., 2015) to predict fontanelle variables for clinical examination, when chronological age is known, or to estimate age‐at‐death in a forensic context. Specifically, the non‐linear model formula for a three‐parameter logistic model is given by:

$$\widehat{\mathrm{Y}}={\beta }_0/\left(1+\exp \left({\beta }_1\mathrm{x}\left[\log \left(x\right)-\log \left({\beta }_2\right)\right]\right)\right)$$

where $\widehat{\mathrm{Y}}$ constitutes the mean response variable, i.e. age (months), β₀ represents the intercept of the value when x = 0, β₁ is the inflexion point and β₂ is the shape of the curve. The nonlinear model framework was then modified to incorporate possible heteroscedasticity by specification of a variance function, which depends on the regression parameter β through the mean function $\widehat{\mathrm{Y}}$, in the form of a power function. This specification is represented as:

$$\text{Var}\left(\widehat{\mathrm{Y}}\right)={\sigma }^2{\widehat{\mathrm{Y}}}^{\theta },$$

where the scale parameter σ governs the overall precision in the response, while the variance parameter θ specifies fully the functional form. This is a common variance function for biological data (Carrol and Rupert, 1988; Davidian and Giltinan, 1995). Modelling the mean relationship yields an accurate estimation of coefficient standard errors, and hence reduces the chance of failing to detect significant effects. An alternative to this approach would be to transform the age variable; however, we opted for variance modelling in order to retain the scale of age in months.

Focusing exclusively on changes in the mean may underestimate, overestimate or fail to distinguish real nonzero changes in heterogeneous distributions. Therefore, quantile non‐linear regression was performed to estimate the conditional quantiles of the response variable distribution in the aforementioned logistic model in order to provide a more complete view of possible causal relationships between variables in growth and development. As there was no evidence of heterogeneity in the coronal diameter, variance function modelling was not required and the relationship between coronal diameter and chronological age was modelled using linear regression. Quantiles corresponding to the 5th and 95th percentile of the distribution (covering 90% of the population distribution) were modelled and coefficients estimated, which is of particular significance in developing growth charts of a population as a screening criterion for abnormal growth.

3. RESULTS

3.1. Temporal patterns of maturation of the anterior fontanelle

Descriptive statistics for each variable of interest with age and sex effects are reported in Table 1. The greatest velocity in the growth of the calvarium occurs within the first 6 months, specifically between the 2‐ to 3‐month and 3–6 month cohorts (p < .1); as well as 3–6 and 6–9 month cohorts (p < .05) for maximum cranial length (GOL) (Figure 2). An increase ≥ 10.13 mm occurs between 3 and 9 months postpartum in GOL and maximum cranial breadth (XCB) in Queensland infants, prior to a plateau in the growth rate of XCB and basio‐bregmatic height (BBH) between 12 and 30 months (≤ 3.14 mm/4‐month interval) in contrast to GOL, which continues to increase in length up to 30 months (Figure 2).

TABLE 1.

Descriptive statistics (mean ± standard deviation) for modern Queensland males and females in each age cohort (months postpartum)

n	0–1		1–2	2–3	3–6	6–9	9–12	12–16	16–20	20‐24	24‐30
	10 | 7		4 | 3	6 | 8	11 | 11	19 | 12	11 | 9	17 | 20	21 | 17	15 | 12	13 | 10
AFSA	♂	2,254.60 ± 1,053.89	1,258.65 ± 884.23bc	1,042.59 ± 412.68cd	1,050.04 ± 547.79	483.21 ± 278.01ef	250.86 ± 202.35h	80.62 ± 108.03h	42.88 ± 69.17h	0.00 ± 0.00h	0.00 ± 0.00
	♀	1954.09 ± 925.24a	2,360.22 ± 1,174.79	1,386.37 ± 828.41c	929.95 ± 247.07d	637.11 ± 334.49efg	321.09 ± 213.48h	187.67 ± 147.55h	67.36 ± 83.21h	8.95 ± 26.86h	0.00 ± 0.00
	–		–	–	–	–	–	<0.01	–	‐	‐
NFSA	♂	22.19 ± 8.09	13.30 ± 10.31bc	9.77 ± 4.57c	8.85 ± 4.87	3.35 ± 2.00h	1.57 ± 1.27h	0.48 ± 0.65h	0.25 ± 0.41h	0.00 ± 0.00	0.00 ± 0.00
	♀	25.03 ± 10.68a	22.64 ± 10.53	12.11 ± 6.81c	7.95 ± 2.07d	4.79 ± 2.49h	2.13 ± 1.41h	1.43 ± 1.08h	0.71 ± 0.96h	0.06 ± 0.18h	0.00 ± 0.00
	–		–	–	–	–	–	<0.01	–	‐	‐
SLD	♂	113.40 ± 5.61ab	100.78 ± 16.23bc	100.75 ± 13.09c	82.73 ± 32.31	28.93 ± 12.43e	18.78 ± 11.74fg	7.35 ± 8.91h	4.63 ± 7.02h	0.00 ± 0.00	0.00 ± 0.00
	♀	109.64 ± 3.49ab	114.94 ± 12.35b	93.47 ± 29.22c	82.65 ± 25.53	41.01 ± 13.53e	22.47 ± 13.45fg	17.34 ± 10.85h	7.83 ± 8.13h	1.00 ± 3.00h	0.00 ± 0.00
	–		–	–	–	<0.05	–	<0.01	–	‐	‐
CLD	♂	80.62 ± 10.52a	63.22 ± 15.47bc	59.77 ± 16.86c	54.29 ± 12.37	39.54 ± 12.31	23.83 ± 16.07f	10.19 ± 10.56 h	5.57 ± 8.41h	0.00 ± 0.00h	0.00 ± 0.00
	♀	77.89 ± 8.20ab	78.01 ± 14.04bc	61.86 ± 20.40cd	55.45 ± 4.06d	48.20 ± 5.02	29.05 ± 10.98f	22.53 ± 12.08g	13.20 ± 13.43h	1.63 ± 4.87h	0.00 ± 0.00
	–		–	–	–	–	–	<0.01	<0.10	‐	‐
GOL	♂	121.28 ± 5.87ab	123.65 ± 6.33bc	130.65 ± 2.29c	139.33 ± 3.78	149.73 ± 7.91e	153.84 ± 10.42f	160.64 ± 9.69h	162.35 ± 6.79h	169.08 ± 7.21h	170.13 ± 9.71
	♀	117.52 ± 4.15ab	120.87 ± 2.55bc	126.59 ± 1.73c	135.25 ± 7.34d	142.49 ± 6.53e	151.19 ± 6.06h	157.33 ± 8.41h	158.37 ± 6.21h	159.73 ± 8.92h	159.17 ± 7.81
	–		–	<0.01	–	<0.05	–	–	<0.10	<0.01	<0.05
XCB	♂	102.95 ± 10.67abc	101.43 ± 5.23bc	109.37 ± 2.00c	110.95 ± 9.21	123.75 ± 7.43e	127.18 ± 5.28f	131.51 ± 7.32h	132.21 ± 5.52h	134.07 ± 9.39h	134.97 ± 7.61
	♀	97.48 ± 6.94 a	107.66 ± 0.84bcd	113.29 ± 4.15cd	111.89 ± 6.57d	117.57 ± 6.83	126.63 ± 4.71h	125.61 ± 5.76h	130.04 ± 7.07h	128.92 ± 2.78h	131.75 ± 4.21
	–		<0.05	<0.10	–	<0.05	–	<0.05	–	‐	‐
BBH	♂	97.85 ± 5.55a–d	96.95 ± 2.37b–e	102.19 ± 3.85c–f	106.03 ± 6.89d–f	109.49 ± 27.77ef	117.40 ± 5.23h	120.15 ± 5.75 h	124.26 ± 5.51h	127.30 ± 5.75 h	126.75 ± 4.68
	♀	90.30 ± 7.63a	102.64 ± 5.01bc	104.10 ± 2.49c	104.63 ± 2.34	114.01 ± 3.35ef	116.13 ± 4.39fg	119.33 ± 3.95h	120.72 ± 3.91h	121.69 ± 4.51h	122.23 ± 2.27
		‐	‐	‐	‐	‐	‐	‐	<0.05	<0.05	<0.05

Abbreviations: AFSA, absolute fontanelle surface area (mm²); BBH, basio‐bregmatic height, determined between superior‐inferior extreme position planes; CLD, maximum coronal diameter of anterior fontanelle (transverse diameter); GOL, modified maximum cranial length (Lottering et al., 2014) between anterior‐posterior extreme position planes; NFSA, normalised fontanelle surface area (%); SLD, maximum sagittal diameter of anterior fontanelle (anterior‐posterior diameter); XCB, maximum cranial breadth.

Unit of measurement: millimetres (mm or mm²). Segregated sample numbers for males and females expressed as M | F.

Statistically significant differences at p < .05 between all age cohorts segregated for sex, except for ^a1–2 months; ^b2–3 months; ^c3–6 months; ^d6–9 months; ^e9–12 months; ^f12–16 months; ^g16–20 months; and ^h all proceeding age cohorts. Sex effects at p < .05 within each age cohort are provided, if present, beneath male and female descriptive statistics, for each variable.

FIGURE 2.

Age‐related growth patterns of maximum cranial length (GOL, circles with solid line), breadth (XCB, squares with dotted line) and basio‐bregmatic height (BBH, diamonds with dashed line) of Australian males from birth to 24 months postpartum, pooled for sex. NOTE: Each co‐ordinate represents the mean for each age subset in months, reported in mm. Error bars represent the standard deviation for each age subset

Representative cranial models, selected based on the median fontanelle dimensions for each age subset, are provided in Figure 3, illustrating temporal maturation of the fontanelle network. In the assessment of the connective tissue area encompassing the anterior fontanelle and contiguous sutures (AFSA, NFSA) and the maximum sagittal and coronal diameters of this tissue network, a consistent decline with postnatal age is evident. As variation in AFSA was most substantial between birth and 6 months of age (95% CI 769.03–2,666.7568 mm²), AFSA was normalised (NFSA) with reference to the cranial dimensions GOL and XCB. The greatest rate of change of 41.59% in NFSA in females occurs between 1 and 3 months postpartum (p < .05), prior to reaching 19.15% of the original area by 6 months of age (Table 1). A significant age‐related decline of 40.05% in NFSA between 0–1 month and 1–2 month cohorts was evident in males (p < .05). After 9 months of age, 92.93% of the fontanelle network has ossified. In the 12–16 month cohort, females exhibit significantly greater (p < .05) remnants of fibrous tissue (5.75%) in contrast to males (2.18%)—the only age cohort to demonstrate significant sexual dimorphism in growth trajectories (p < .05), coinciding with earlier AF closure in males.

FIGURE 3.

Isosurface models demonstrating temporal closure of the anterior fontanelle and contiguous sutures in Australian infants aged from birth to 20 months from a superior perspective. Representative crania based on median normalised fontanelle surface area (NFSA) for each age cohort. Note wide contiguous sutures from birth to 1 month and delayed closure of the sagittal suture. Data are cross‐sectional; each image is sourced from a different individual; images are not of equal scale

Temporal decline in sagittal diameter is most significant between the 3–6 and 6–9 month cohorts (p < .05), independent of sex. A significant decline in coronal diameter occurs 3 months later between the 6–9 and 9–12 month cohorts (p < .05) (Table 1).

3.2. Sexual dimorphism

Generally, females exhibit larger linear and areal dimensions of the fontanelles than males for absolute and standardised variables (Table 1) from birth to 30 months. Statistically significant sexual dimorphism was evident in the 12–16 month cohort (p < .05) for all fontanelle variables, an age milestone consistent with the earlier closure of the anterior fontanelle in males. In males, a significant decline in NFSA and with the 3–6 and 9–12 month cohorts for NFSA in females (p < .05), approximately 3 months later. For GOL, XCB and BBH, sexual dimorphism was a frequent phenomenon, with males exhibiting greater dimensions than females across the majority of the age subsets, with the exception of individuals ≤ 6 months for XCB and BBH (Table 1). Although these assertions of sexual dimorphism provide an insightful into the differential growth and development of Queensland boys and girls, the mean variation constitutes < 1 cm difference for most subsets.

3.3. Directionality of closure

Spearman’s correlation coefficients demonstrate a strong positive correlation between maximum sagittal and coronal diameters of the anterior fontanelle (r = .98 for the pooled sample), implying that generally the fontanelle network closes in a uniform manner. Visual representations of the fontanelle network for the ‘median’ individual in each age subset (Figure 3) illustrate that the metopic and coronal sutures are the first to close, ossifying relatively synchronously with age (Figure 3). Within the first 6 months, the sagittal diameter is significantly (p < .05) longer than the coronal suture, before the dimensions reach equal proportions in the 6‐ to 9‐month subset, representing the classical diamond morphology. Closure of the sagittal suture appears delayed, with only 27.20% ossification occurring between birth and 6 months, in a posteroanterior direction, prior to a statistically significant increase of 58.72% in the following 3 months (6–9 months) (Figure 4), prior to attaining 95% sutural closure at 16 months. The left and right sides of the coronal suture are the greatest contributors of the persistent anterior fontanelle ≥ 20 months postpartum, demonstrating 87.48% ossification, in contrast to 90.32% of the metopic suture.

FIGURE 4.

Percentage and timing of closure of the metopic, coronal and sagittal sutures, pooled for sex. The greatest rate of ossification of the fibrous tissue of the metopic (solid) and coronal (dotted) sutures occurs within the first 3 months postpartum, occurring in a simultaneous manner; in contrast to statistically significant closure (p < .05) of the sagittal suture (dashed) between 2‐ to 3‐month and 3–6 month cohorts. Error bars represent the standard deviation of the mean for each age subset in months

3.4. Prediction models for clinical examination of age

For assessing developmental patterns of infants of known chronological age, paediatric examination or assessment of infants victim to suspected maltreatment, non‐linear regression with variance modelling and quantile linear regression models were fit to the data to predict the response variable. Overall sex effects were deemed to be non‐significant for NFSA (p > .05), SLD (p > .05) and CLD (p > .05), therefore sex‐pooled logistic models are appropriate for developmental assessment for the given age range (Tables 2 and 3, Figure 4). NFSA and diameters were divided by 100 for computational purposes and should be transformed into the original units of measurement for assessment. Figure 5 illustrates the logistic model to estimate mean NFSA and quantile regression models at the 5th and 95th percentile fit to the dataset. In clinical examination, individuals that exhibit a response value that exceeds the 90% distribution of the Queensland population may flag up cause for concern. For example, if an infant of 4.3 months presents for cranial examination, based on the models in Table 2, the mean NFSA for this age should be 7.23% (0.232/(1 + exp(1.445*[log(4.3) – log(2.487)])) with the 5th and 95th percentile of the Queensland population exhibiting an NFSA of 1.72% and 16.47%, respectively. The actual NFSA for the infant of interest is 23.85%, which does not fall within the 90% distribution of the subpopulation (Figure 5; asterisk), suggesting possible delayed development, and thus may require further investigation. Figure 6 demonstrates that the mean variance relationship is more constant for the assessment of the coronal diameter compared with the sagittal diameter, implying that the quadratic model utilised is a better fit to the data and thus may be more accurate in age estimation and clinical assessment of infants. Based on the 95th quantile model for the sagittal diameter (Figure 6a), it is evident that ‘gaps’ are present between the ages of 5 and 10 months, with the model interpolating between points, as opposed to fitting the current data points.

TABLE 2.

Quantile non‐linear regression coefficients and standard errors (SE) for the mean normalised fontanelle surface area (NFSA) and sagittal diameter (SLD), with 5% and 95% quantile models, for clinical examination of Australian infants of known chronological age, pooled for sex

(NFSA/100) = (β 0/(1 + exp(β 1*[log(Age(Months)) – log(β 2)]))
NFSA	Mean NFSA		5% Quantile		95% Quantile
	Est.	SE	Est.	SE	Est.	SE
β 0	0.232	0.015	0.106	0.049	0.341	0.021
β 1	1.445	0.165	1.579	0.475	1.649	0.258
β 2	2.487	0.352	1.521	1.193	4.129	0.997

(SLD/100) = (β 0/(1 + exp(β 1*[log(Age(Months)) – log(β 2)]))
SLD	Mean SLD		5% Quantile		95% Quantile
	Est.	S.E.	Est.	S.E.	Est.	S.E.
β 0	1.139	0.038	1.073	0.113	1.335	0.123
β 1	2.093	0.175	2.304	0.255	2.057	0.558
β 2	5.600	0.327	3.188	0.395	9.223	2.093

TABLE 3.

Coefficients for the linear regression with squared term model, for the mean coronal diameter (CLD), with 5% and 95% quantile regression functions, for clinical examination of Australian infants of known chronological age, pooled for sex

(CLD/100) = β 0+β 1+β2x2
CLD	Mean CLD		5% Quantile			95% Quantile
	Est.	S.E.	Est.	Lbd	Ubd	Est.	Lbd	Ubd
β 0	0.812	0.025	0.605	0.576	0.642	1.060	0.952	1.367
β 1	–0.065	0.006	–0.067	–0.075	–0.065	–0.073	–0.097	–0.053
β 2	0.001	0.0003	0.001	0.001	0.002	0.001	0.0008	0.002

FIGURE 5.

Temporal decline in NFSA for Queensland infants aged 24 months and under, pooled for sex, for clinical examination of growth and development of the anterior fontanelle. Quantile regression models at the 5^th and 95th percentiles are represented by dashed lines, and the mean percentage fontanelle (50th quantile) is depicted by the solid line, modelled using a logistic regression with variance. NOTE: Asterisk refers to clinical outlier discussed in Results (‘Predictive models for clinical examination and medico‐legal application’)

FIGURE 6.

Temporal decline in sagittal (anteroposterior) (a) and coronal (transverse) (b) diameters of the anterior fontanelle for utility in clinical examination of Queensland infants aged 24 months and under. Quantile regression models at the 5th and 95th percentiles are represented by dashed lines, and the mean diameter (50th quantile) is depicted by the solid line for (TOP) sagittal diameter (distance between the most antero‐inferior point on the metopic suture to the posterior point on the sagittal suture) and (BOTTOM) the coronal diameter (distance between lateral, unfused points of the coronal sutures)

4. DISCUSSION

In light of considerable population variation in anterior fontanelle (AF) size and closure in infants, the establishment of normative values for surface area (AFSA) and diameters provides contemporary Australian reference data that may enhance evaluation of growth and development between birth and 2 years of age. Although our data acquisition may be specific to MSCT in the present study, we do not advocate prospective MSCT scanning of infants for standard fontanelle examination. Rather, the protocol published by Lottering et al. (2014) may be extrapolated to other 3D datasets acquired by non‐iodising radiation including but not limited to laser/surface scanning, magnetic resonance imaging (MRI) and ultrasonography.

To overcome the assumption of geometric regularity in conventional AF methodologies (Davies et al., 1975; Hall et al., 2006), Moffett and Aldridge (2014) proposed a new method to quantify the surface area of the AF in amira® on MSCT scans of skulls in the Bosma Collection based on the radiodensity of connective tissue via voxel selection. Similarly, Kirkpatrick et al. (2019) quantified the surface area of the AF in a non‐accidental MSCT sample of New Zealand infants using 3D oblique multi‐planar reformatting features in OsiriX®. Using a CAD workflow, Lottering et al. (2014) proposed an automated areal quantification method in geomagic design X^TM based on the interpolation of coordinates using the reference geometry of the isosurface. Moffett and Aldridge (2014) report that 15.30% variance in AFSA is attributed to the measurement error between the conventional (Smith and Popich, 1972) and 3D areal protocols, demonstrating that the traditional method is more likely to overestimate AFSA. It is important to reinforce that the quantification of the fibrous membrane for individuals < 6 months postpartum in our study is not restricted to the AF, but includes the contiguous sutures and fontanelles (Figure 3), thus direct comparison with studies employing the conventional methodology to assess AFSA (Ogunye et al., 1982; Chakrabarti, 1989; Mattur et al., 1994; Kirkpatrick et al., 2019) is not possible below this age.

Using MSCT volume‐rendered reconstructions, Pindrik et al. (2014) provided normative ranges for AF surface area for US infants aged from birth to 2 years, based on the Davies et al. (1975) measurement protocol. For comparative purposes, descriptive statistics of AFSA in the Australian infants were re‐calculated with reference to The American Academy of Pediatrics recommended clinical age cohorts (7–9.9 , 10–12.9, 13–15.9, 16–18.9, 19–21.9 and 22–24.9 months) and demonstrate that for all age‐matched cohorts the median AF surface area of Australian children is substantially larger than the US children. As these comparisons refer to absolute fontanelle dimensions, we anticipate that the larger Australian dimensions may be attributed to possible secular change in cranial vault dimensions of children in the current milieu, in comparison with infants born in the 1970s and 1980s (Popich and Smith, 1972; Duc and Largo, 1986; Kataria et al., 1988). Significant secular change in craniofacial form has been documented in US adults from the mid‐19th century to the 1970s (Jantz and Meadows Jantz, 2000), the authors suggesting that the principal effects responsible for increases in vault height and decrease in maximum cranial breadth, occur early in the growth period (Alberman et al., 1991). The present study may provide preliminary evidence for this proclamation, based on AF dimensions.

Measurement of linear AF size as per definitions by Popich and Smith (1972) (mean of the SLD and CLD) demonstrate significantly larger dimensions compared with infants born between 1974 and 1978 in the Second Zurich Longitudinal Study (Duc and Largo, 1986). Specifically, between 9 and 12 months the mean fontanelle size in Swiss males and females is 14.2 ± 6.4 and 14.9 ± 7.3 mm (Duc and Largo, 1986), in contrast to 36.36 ± 5.07 and 43.27 ± 7.15 mm in our sample. Between 12 and 18 months, Australian individuals exhibit a larger AF size (16.64 ± 2.73 mm) in comparison with Chinese (10.2–12.7 mm) (Chang and Hung, 1990), Nigerian (8.0 ± 10.0 mm) (Omotade et al., 1995) and US caucasian infants (6.0 ± 6.0 mm) (Kataria et al., 1988).

Modelling of NFSA demonstrates that intra‐cohort variation is greatest within the first 3 months postpartum, as illustrated by the funnel nature of the quartile curves (Figure 4). This variability is congruent with the literature on full‐term infants (Duc and Largo, 1986; Moffett and Aldridge, 2014) and may be attributable to differences in the initial location of the primary ossification centres, differences in intracranial growth rates during gestation or cranial deformations during parturition. The normalisation of AFSA also demonstrates a decrease in NFSA within the first 2 months, suggesting that the increase in absolute area in Queensland females is correlated to variable head size at this age. This may be attributable to intrinsic genetic variation; the results are consistent with the peak in median AFSA around 1–2 months of age reported in clinical literature (Duc and Largo, 1986; Pedroso et al., 2008; Pindrik et al., 2014); however, our study is the first to employ a transformation technique to normalise the size of the cranium. The positive age‐related changes in anthropometric dimensions of the cranium illustrated in Figure 2 exemplify the importance of removing the effect of size when assessing volumetric and areal changes of the cranium in infants < 3 months of age.

Pooled for sex, growth of the neurocranium is most significant within the first 8 months postpartum, with an increase of 19.03 mm in GOL and 10.04 mm in XCB (Table 1), prior to slowing after the infant’s first birthday. Between 12 and 24 months, maximum cranial length increases at a rate 3.72× that of XCB and 5.57× that of BBH, with an increase in the dimension of vault height (BBH) delayed by 3 months (p < .05), in contrast to the neurocranium. Significant sexual dimorphism was observed in the cranial dimensions GOL, XCB and BBH as early as 1–2 months, which would indicate that sex differences should be present in AFSA. However, this was not the case in the present study, with the exception of females exhibiting significantly greater areal dimensions (AFSA and NFSA) than males in the 12–16 month age cohort (p < .05) only. Despite the statistical significance reported for a handful of cohorts in this study, overall sex effects in the non‐linear variance models are not significant for increasing chronological age, which supports the studies conducted on Nigerian (Ogunye et al., 1982; Adeyemo et al., 1999; Uzukwu‐Edeani et al., 2013) and European infants (Popich and Smith, 1972; Davies et al., 1975).

Upon treating the contiguous sutures as separate entities, as evidenced by the distance curves for the percentage closure of the sagittal, coronal and metopic sutures (Figure 4), our results demonstrate that the pattern and velocity of closure of the sagittal suture significantly deviates from the rest of the fontanelle network. Less than 5% of the sagittal suture has closed before 2 months postpartum followed by a significant catch‐up maturation event between 3–6 and 6–9 month cohorts with a mean increase of 58.7% (p < .05). This maturation event appears intrinsically correlated with the peak velocity in cranial growth in the sagittal plane (GOL) between these ages, possibly attributable to an increase in brain volume. In contrast, the rate of closure of the metopic and coronal sutures is highly correlated from birth. Craniosynostosis of the sagittal suture diagnosed in infants may have more profound functional implications for brain development than would premature closure of the coronal or metopic sutures.

Point‐to‐point measurement of the fontanelle’s extremities and respective sutures, which constitute the diameters, is comparatively quicker than the areal measurement and may be preferred in clinical assessment (Tables 2 and 3). Diameter standards may also be more applicable when employing MRI or ultrasonography, where two‐dimensional measurements are easier to obtain if volume data are unavailable. Overall, we recommend that a multi‐variate approach to paediatric assessment and/or age estimation be employed, using all three prediction functions where possible, in an attempt to increase the precision of the response variable. The normative standards provided are more representative and suitable for bone age assessments and estimation of the current milieu, overcoming the temporality and demographic biases inherent in current standards when applied to Australian infants. It should be recognised that our paediatric sample, acquired from predominantly metropolitan Brisbane children’s hospitals, may result in selection bias; however, the characteristics are unlikely to deviate significantly from the general paediatric population, as most individuals were selected on the basis of exhibiting ‘normal skeletal findings’ on the radiology report. A limitation of the current study is the exclusion of Australian Indigenous (Aboriginal and/or Torres Strait Islander) individuals, which represent the population minority. Our statistical projections estimated that a robust sample size of modern Indigenous children for each age cohort would not be attainable from the target hospitals within the data collection period; and we also hypothesise that inclusion of this ancestral group may present as statistical outliers due to suspected differences in socio‐economic status and/or a predisposition to advanced maturation (Garn and Bailey, 1978; Cameron et al., 1993; Bogin, 1999). This statement may be supported by a recent New Zealand study by Kirkpatrick et al. (2019). For example, the mean AF surface area (pooled for sex) for Australian infants aged 6–9 months is similar to New Zealand European infants (Australian: 639 mm², NZ European: 650 mm²), but 41% larger than Maori/Pacific infants in the same age cohort. This is corroborated by Australian infants exhibiting larger surface areas between 10 and 12 months (Australian: 285.9 mm², NZ Maori/Pacific: 185 mm²). We recommend that future work on cranial growth in the Queensland population utilises the current dataset as the basis for an informative prior while incorporating a Bayesian modelling approach; this was not possible in this study, as a suitable informative prior distribution was not available. Future work to validate the proposed models on Australian Aboriginal and/or Torres Strait Islander children is recommended prior to clinical application on minority groups.

AUTHOR CONTRIBUTIONS

N.L. designed the study, conducted all data collection and quantitative analysis, convened the authorship team and had final editorial decision over the content of the submission. N.L. and M.B. designed the protocol and 3D methodological approach in CAD software. C.A. was responsible for statistical analysis, modelling and data interpretation. N.L., M.B., L.G., D.M. and C.A. actively participated in development of ideas in this paper and were part of the core authorship team, adding text, copy‐editing, providing resources and feedback throughout the review of the manuscript.

Lottering N, Alston CL, Barry MD, MacGregor DM, Gregory LS. Temporal mapping of the closure of the anterior fontanelle and contiguous sutures using computed tomography, in silico models of modern infants. J Anat. 2020;237:379–390. 10.1111/joa.13200