SEARCH

SEARCH BY CITATION

Keywords:

  • craniofacial development;
  • computed tomography;
  • human evolution;
  • fontanelle

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

The anterior fontanelle (AF) is an integral element of the developing human infant craniofacial system. Consideration of the AF is crucial for assessing craniofacial growth, as altered development of this feature may indicate abnormal growth. Moreover, prolonged patency of the AF may represent a derived hominin feature. The AF is regarded as essential for fetal head molding during birth in humans, with deformation of the head during birth often necessary for successful delivery. However, the function of a patent AF among fossil hominins is unclear. Because the AF represents an important structure in both a clinical and evolutionary context, techniques for estimating the size of the AF must be accurate and reproducible. Therefore, we have developed a novel method for assessing surface area of the AF with the goal of creating a more accurate measure of this feature. In this study, we test the accuracy and repeatability of a novel three-dimensional (3D) method for assessing the size of the AF in human infants and compare the results obtained for surface area of the AF using the conventional and 3D methods. Anat Rec, 297:234–239, 2014. © 2013 Wiley Periodicals, Inc.

The human infant skull is highly malleable relative to the adult condition, with its bones separated by unfused sutures and fontanelles. The anterior fontanelle (AF) is a curved rhomboid, nonmineralized fibrous membrane in the cranial vault at the convergence of the coronal, sagittal, and metopic sutures in the developing fetus and infant (Fig. 1). Rather than being a structure in the traditional sense, the AF is actually the residual membranous remains of the ectomeninx, the neural crest cell-derived tissue from which the calvarial vault bones are formed (Jiang et al., 2002). It has been hypothesized that the structure of the AF may reflect the rate of ossification of the bones of the calvaria, the relative placement of those ossification centers, the rate of increase in intracranial volume, or some combination of these factors (Dechant et al., 1999; Mathijssen et al., 1999). Closure of the AF in humans modally occurs prior to 2 years of age (Aisenson, 1950; Acheson and Jefferson, 1954; Duc and Largo, 1986). In contrast, this structure is obliterated much earlier in nonhuman primates; at the time of birth the AF is generally closed in monkeys and quite small in apes (Schultz, 1936, 1969; Dolan, 1971; Abitbol, 1993). Moreover, a juvenile specimen of Homo erectus (Perning 1) estimated to be 1 year old (Coqueugniot et al., 2004) as well as a specimen of Australopithecus africanus (Taung) estimated at 3–4 years old (Dart, 1925; Lacruz et al., 2005) both preserve remnants of the AF. Thus, a prolonged period of patency of the AF and related metopic suture may represent a derived anatomical feature among hominins compared to other primates (Falk et al., 2012).

image

Figure 1. Schematic illustration of the anterior fontanelle (AF). From left to right: the AF in a normally-developing neonate; a neonate with metopic craniosynostosis; a neonate with Apert syndrome.

Download figure to PowerPoint

Several hypotheses for the persistent patency of the AF in humans have been offered. One of these hypotheses contends that the persistent presence of the AF, combined with unfused sagittal, coronal, and metopic sutures of the cranial vault, functions to allow cranial deformation in the birth canal. The potential for deformation of the cranium during birth would help to solve the “obstetric dilemma” posed by the competing pressures of bipedal locomotion narrowing the pelvis, effectively making the birth canal smaller (Washburn, 1950, 1960; Campbell, 1966; Pinkerton, 1973; McHenry, 1975; Zihlman, 1978; but see Dunsworth et al., 2012), and increasing encephalization in the newborn (Washburn, 1960; Lindburg, 1982; Tague and Lovejoy, 1986). Thus, these competing constraints imposed by bipedality and brain size evolution in the human lineage may have necessitated a modification in frontal neurocranial ossification (Falk et al., 2012), including retention of the AF. However, this hypothesis does not explain why other primates, including Saimiri sciureus, whose birth canal is even more constrained relative to neonatal head size than that of humans (Tague and Lovejoy, 1986), do not retain the AF through birth (Dolan, 1971). An alternative hypothesis is that prolonged retention of the AF and associated sutures could result from either overall or localized rapid brain growth in humans during the fetal and infant stages during which the growth of the underlying neural tissue outpaces that of the overlying bone (Martin, 1983; Falk et al., 2012). Whatever the explanation for retention of the AF in human infants, the fact that a postnatally large and persistent AF only appears in hominins suggests that it may represent a derived feature.

The relative size of the AF is also significant in the modern clinical setting (Fig. 1). Assessment of the AF surface area (AFSA) is routine in infant (Duc and Largo, 1986) and fetal examination as three-dimensional (3D) ultrasound techniques have made in utero assessment of craniofacial morphology, including the AF, possible (Chaoui et al., 2005; Faro et al., 2005, 2006). Such assessment is critical as irregular development in this structure may indicate abnormal craniofacial growth (Davies et al., 1975; Philip, 1978; Paladini et al., 2007, 2008). For example, reduction of AFSA is common among premature infants with intrauterine growth retardation (Davies et al., 1975; Philip, 1978) and fetuses with craniosynostosis (Kreiborg et al., 1993; Cohen and MacLean, 2000), whereas large fontanelles (including the AF) are common among fetuses with Down syndrome (Paladini et al., 2007, 2008) and congenital hypothyroidism (Smith and Popich, 1972). Moreover, irregular AFSA may indicate cardiac overload and central nervous system malformations (Paladini et al., 2008).

Because the AF is such an integral component of the developing craniofacial system, a basic, common technique has been employed for its measurement. The conventional method for measuring AFSA consists of measuring the maximum anteroposterior (AP) and mediolateral (ML) linear lengths of the AF (Dubowitz et al., 1970; Popich and Smith, 1972; Malas and Sulak, 2000), modeling the AF as a consistently rhomboidal shape (Fig. 2A). However, this model does not reflect the real shape of the AF as a curvilinear structure, nor can it account for the naturally varying shape of the AF both within and among species (Fig. 1). For example, bregmatic fontanelle bones commonly appear during intrauterine development in some primates, but are less common in humans (Schultz, 1923), whereas, within humans, the size and shape of the AF is variable even during normal development (Popich and Smith, 1972; Faix, 1982; Kiesler and Ricer, 2003). Moreover, the conventional technique reduces the 3D nature of the AF, which may span a large, curved surface, to two dimensions (2D), though it is unclear whether this reduction of dimensionality affects the accuracy of the measurement.

image

Figure 2. (A) Conventional method for estimating anterior fontanelle surface area (Dubowitz et al., 1970), where AB: maximum anteroposterior dimension of the fontanelle; CD: maximum mediolateral dimension of the fontanelle. Illustrated on three-dimensioal (3D) reconstruction of CT data from Bosma Collection specimen 16. Two-dimensional measurements of the anteroposterior length (AB) and mediolateral width (CD) of the anterior fontanelle are shown in green lines. The shape being estimated by the measurements in green is outlined in gray. (B) Voxel selection on coronal and sagittal CT slices, and the resulting 3D SurfaceGen illustrated with CT data from Bosma Collection Specimen 300. Voxels labeled as bone are illustrated in purple and the fibrous covering of the AF in green.

Download figure to PowerPoint

Given the evolutionary and clinical significance of the AF, techniques for its measurement must be accurate, repeatable, and representative of its true nature. Thus, the purpose of this study is to test a new technique, which uses 3D data for estimation of AFSA in human infants and to compare the accuracy of these measurements with those obtained using the conventional 2D technique.

Materials and Methods

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

Study Sample

The study sample consisted of computed tomography (CT) images of museum-curated infant crania (n = 5) from the Bosma Collection (Shapiro and Richtsmeier, 1997). These specimens were acquired by James F. Bosma, M.D., from the State of Maryland Anatomy Board in the 1950s, were aged using sequential dental eruption patterns and tooth bud formation, and are currently housed at Pennsylvania State University; the cause of death for these specimens is unknown, though none show any craniofacial anomalies (Shapiro and Richtsmeier, 1997). Because the AF is modally measured during late pregnancy or shortly after birth in order to assess craniofacial growth (Duc and Largo, 1986; Chaoui et al., 2005; Faro et al., 2005, 2006), the infants included in this study range from 30 to 36 weeks gestation as determined by Shapiro and Richtsmeier (1997). CT scans were performed at Johns Hopkins Medical Institutions on a GE Medical Systems Genesis Jupiter scanner, 512 × 512 matrix, 0.39-mm pixel size, and 1.5-mm slice thickness. All crania were scanned using the same imaging protocols.

Data collection & analysis

For each of the five infants, 3D surfaces of the cranium were reconstructed from the CT images using Amira 5.2© software. Bony tissue was selected within each slice (using Amira's “Magic Wand” tool), from which a 3D surface image of the bony cranium was created (SurfaceGen).

Using the conventional method, AFSA is estimated via 2D measures of maximum AP length and ML width (Fig. 2A), where area = (AP*ML)/2 (Dubowitz et al., 1970; Popich and Smith, 1972; Davies et al., 1975; Malas and Sulak, 2000). Maximum AP and ML dimensions of the AF were measured (“3D Length” tool) on the 3D image of each skull (Fig. 2A). Linear measurements were collected three times on each skull to calculate three AFSA values.

In the proposed 3D method, the membranous layer of the AF is defined in the CT images separate from bone (Fig. 2B). Voxels along the membranous AF were manually selected in each CT slice (“Brush” tool); this was used so that only the voxels along the top of the membranous AF were selected, thus minimizing overestimation of AFSA. Voxel selection was performed in both coronal and sagittal planes in order to ensure a continuous layer along the AF surface. A 3D image was again created for each cranium, and a smoothing function was employed (i.e. “Constrained Smoothing”), such that the AF surface was smoothed without changing designated voxel labels. Finally, the voxels selected on the external surface of the AF were used to calculate AFSA (“Calculate Surface Area” tool). These values were then divided by two, as the surface area calculated with this algorithm includes both the superior and inferior surfaces of the AF. This process was repeated three times for each skull to calculate three AFSA values as with the conventional method. Mean AFSA values were calculated for each individual using each method (Table 1).

Table 1. Estimated anterior fontanelle surface areas using both the conventional and 3D methods
SpecimenTrial 1 AFSATrial 2 AFSATrial 3 AFSAMeanMean deviation (%)
Conventional method
Cranium 1307.0310.7310.4309.40.5
Cranium 2439.0424.6437.6433.71.4
Cranium 31,118.01,148.71,109.01,125.21.4
Cranium 41,124.41,136.21,126.21,128.90.4
Cranium 52,227.82,175.82,115.62,173.01.8
Three-dimensional method
Cranium 1397.5381.5375.0384. 72.2
Cranium 2334.0371.0375.5360.24.8
Cranium 3873.5884.0871.5876.30.6
Cranium 4613.9595.5594.5601.31.4
Cranium 51,318.51,269.01,278.01,288.51.5

Multiple random and confounding factors may introduce variation into the AFSA estimation, including individual variation, differences between the methods of estimation, and observer error. Therefore, a nested ANOVA procedure was used to determine the proportion of variance due to each of these confounding variables following Kohn and Cheverud (1992) and Aldridge et al. (2005). Using a general linear model (SYSAT 11©), mean squared error values were used to determine the proportion of variance in AFSA due to (1) differences between individual skulls, (2) differences between the 3D and conventional methods, and (3) intrarater error. The proportion of total variance due to differences between individual skulls represents between-subject difference. The proportion of total variance due to differences between AFSA values for the same individual (within-subject variance) represents error due to measurement technique. Finally, the proportion of total variance due to the residual represents the error due to observer, or intrarater error.

Repeatability is a measure of precision relative to the magnitude of difference between individuals (Kohn and Cheverud, 1992). Statistical significance of the ratio of between-subject variance to within-subject variance was evaluated as the intraclass correlation following Aldridge et al. (2005). Intrarater error within each method was also assessed as percent difference between repeated trials for each skull following Collins et al. (1995). The percent difference between AFSA calculated for each trial and the mean AFSA estimated by each method was calculated as:

  • display math

where δ is the percent difference in AF surface area, AFSAMx is the mean AFSA value for a given individual estimated for each method (x), and AFSAI is the AFSA value calculated in a single trial. Average deviation for each method is the calculated as the sum of all δ values for each method divided by the total number of trials (i.e. 15, or three trials per method for each of the five crania).

Results

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

Total variance in AFSA is parsed into three sources: (1) variance due to individual differences, (2) variance due to estimation method, and (3) intrarater error. Results of the nested ANOVA (Table 2) show that the variance in AFSA due to individual infant differences comprises the largest proportion of total variance, 84.68%. In contrast, 15.30% of the total variance is due to the method used, while variance due to intrarater error contributes 0.02% of the total variance.

Table 2. Sources of error and associated percent of total variance in AFSA
Types of varianceMean squared errorPercent variance in AFSA (%)
Between-subject error (differences between individuals)1,883,019.6984.68
Within-subject error (measurement technique)340,166.5415.30
Intrarater error (observer error)520.610.02
Total2,223,706.84100.00

We calculated repeatability for estimating AFSA using the two techniques and compared the methods (Table 3). Measures of intrarater error were similar in the two methods. Average percent deviation for individual trials from the mean for each AF was 1.1% using the conventional method, compared to 2.1% using the 3D method. Modally, larger values were obtained for AFSA using the conventional method (Table 3). Only one of the five individuals showed larger estimates of AFSA calculated with the 3D method than with the conventional method. Intraclass correlations also showed both the conventional (ICC = 0.999) and 3D (ICC = 0.998) methods to be highly repeatable.

Table 3. Repeatability of the two methods for estimating anterior fontanelle surface area (AFSA)
 Mean δa (%)Intraclass correlation coefficient
  1. a

    Mean δ = Σ{[(AFSAMean − AFSAIndividual)/AFSAMean] × 100.0%}/n, where n = the number of individuals.

Conventional method1.10.999
3D method2.10.998

Discussion

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

The large proportion of variance due to individual differences in AFSA indicates that the size of the AF is highly variable in human neonates, even within our small sample. It is possible that the high variability in AFSA may be partly due to the slight differences in ontogenetic ages of the neonates included in this study. However, the high variability in AFSA we find in this study is congruent with previous research on term infants (Duc and Largo, 1986). Thus, it is possible that this high variability in AFSA is not merely due to either the small sample size or the slight variability in ontogenetic age of the present study sample, but is rather characteristic of late-term infant AFSA. If high variability in AFSA is indeed characteristic of term neonates, such variability may be due to slight differences in the initial location of the primary ossification centers of skull, to differences in intracranial growth during gestation, or to cranial deformation during birth. However, our finding of high variability in interindividual AFSA should be confirmed with a larger number of individuals in future studies. Additionally, future studies should address whether high variability in AFSA also characterizes late second and early third trimester fetuses. Moreover, future studies should address whether interindividual variability in AFSA is reduced when AFSA is scaled to other traditional somatic growth measures, such as sitting height or femoral length, although such scaling was impossible in the present study due to the nature of the study sample.

A substantial proportion of the variance is also attributable to differences in the method used to estimate AFSA. Modally, the mean AFSA values calculated with the 3D method are smaller than those obtained using the conventional method. Thus, the results of this study suggest that the conventional method may be more likely to overestimate AFSA. Moreover, the fact that the conventional method tends to produce larger values for AFSA makes it unlikely that the differences in AFSA obtained using the two methods can be attributed to the reduction of the AF to a 2D structure. More likely, the difference between the two methods can be attributed to the assumption inherent in the conventional method that the AF is an idealized, or uniform, rhomboid shape. The conventional method does not take into account the naturally curved borders of the AF, which vary greatly among individuals (Fig. 3). Thus, we believe that the 3D method used in this study represents a more accurate measure of AFSA.

image

Figure 3. Illustration of variation in the shape and size of the AF in the five neonate crania in this study as represented by 3D reconstructions of CT images (not to scale). From left to right: Bosma Collection specimens 6, 16, 300, 402, and 407.

Download figure to PowerPoint

Although our study used CT data, it is important to note that the 3D technique we describe here may be employed with other types of 3D data in addition to CT, including 3D stereophotogrammetry, laser surface scans, magnetic resonance imaging (MRI), and 3D ultrasound data. Thus, the 3D technique represents a method useful in both clinical and research settings when the appropriate data are available.

Multiple developmental anomalies are associated with irregular craniofacial growth and development, including the growth of the AF. Indeed, irregular prenatal development often results in the true shape of the AF being far from the rhomboid shape that is assumed using the conventional technique (Fig. 2A). Therefore, the 3D technique we describe in this study may represent a more accurate measure when evaluating AFSA in such individuals. Ultimately, greater accuracy in AFSA estimation could lead to more accurate clinical diagnoses in utero.

Similarly, the 3D method for evaluating AFSA can incorporate bregmatic fontanelle bones in the estimation, which occur frequently in other mammalian species (Schultz, 1923). Thus, our results suggest that the 3D method may also be more suitable for examining differences in AFSA among species, allowing further examination of interspecies variation in development and subsequent closure of the AF. In particular, recent research suggests modification of the growth patterns of the frontal neurocranium among hominins (Bookstein et al., 1999; Rosenberger and Pagano, 2008; Athreya, 2012; Falk et al., 2012; Freidline et al., 2012a, 2012b; Gunz and Bulygina, 2012; Tague, 2012). Thus, future studies of comparative AF development and morphology may shed light on the evolution of the uniquely hominin trait of a persistently patent AF at birth and in early infancy.

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

The authors thank Dr Joan Richtsmeier for access to the CT data from the Bosma Collection. They also thank Dr Kevin Middleton for assistance with statistics as well as Ian George for technical assistance with Amira. Two anonymous reviewers provided thoughtful comments on the manuscript.

LITERATURE CITED

  1. Top of page
  2. ABSTRACT
  3. Materials and Methods
  4. Results
  5. Discussion
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED