Morphometric Analysis of the Condylar Part of the Occiput Using 3D Computed Tomography Models for Sex Estimation in a Korean Population

Seok-Ju Choi; Yi-Suk Kim

doi:10.7580/kjlm.2026.50.2.53

Abstract

This study aimed to analyze the 3D morphology of the condylar part of the occiput in a Korean population and evaluate its applicability for sex estimation. A total of 87 individuals (44 males and 43 females) were included, and measurements of the occipital condyle and foramen magnum were obtained using computed tomography-based 3D reconstructions. Intra-observer reliability was assessed, and univariate and multivariate discriminant analyses were performed to evaluate the discriminative power of each variable. Most measurements showed significant sexual dimorphism, with males generally exhibiting higher values than females. Particularly, variables related to the occipital condyle, including length, width, and area, demonstrated relatively high discriminative performance, whereas those related to the foramen magnum showed comparatively lower discriminative ability. Multivariate analysis further improved the classification accuracy, indicating that the combined use of multiple variables enhanced the ability to distinguish between sexes. These findings suggest that this approach may be particularly valuable in forensic contexts where skeletal remains are incomplete and offer a reliable alternative when more commonly used elements are unavailable.

Key words: Sex estimation · Condylar part of the occiput · Morphometric analysis · 3D computed tomography · Forensic anthropology

Introduction

Sex estimation from human remains is a fundamental component of biological profiling for personal identification and is one of the most critical analyses in forensic anthropology and archaeology [1]. Traditionally, the skull and pelvis have been regarded as the most reliable skeletal elements for sex determination owing to their pronounced sexual dimorphism [2]. However, in actual forensic contexts, these key skeletal elements are often not fully preserved, and remains are frequently recovered in a fragmented state because of trauma or environmental factors [3]. Consequently, there is an increasing need for methods that enable sex estimation using limited skeletal fragments [4].

In particular, the skull base is relatively well protected owing to its anatomical location and is therefore considered a region with high analytical potential, even in partially damaged remains [5]. Among these structures, the condylar part of the occiput, encompassing the occipital condyle and articulating with the first cervical vertebra, and the foramen magnum, which forms the transitional boundary between the intracranial cavity and vertebral canal, exhibit well-defined morphological boundaries, thereby facilitating reliable measurements for sex estimation [6]. Furthermore, this region is subjected to sex-related differences in mechanical loading, and variations in postural, physiological, and motion responses may influence geometric similarities, thereby manifesting as anatomical differences [7,8]. Meanwhile, several studies have reported sex estimation based on the occipital condyle and the foramen magnum within the condylar part of the occiput [9-11]; however, systematic investigations incorporating parameters such as the area of each structure remain limited, and studies focusing specifically on Korean populations are scarce.

In forensic anthropology, biological profiling, including sex estimation, should be conducted with careful consideration of population-specific morphological characteristics [12]. Even within the same anatomical structure, the distribution of morphology and measurements may vary across populations, reflecting differences in growth and development, environmental conditions, lifestyle, and genetic background [13]. Therefore, in the absence of population-specific data on the condylar part of the occiput in Koreans, establishing the morphological characteristics and metric criteria for this area may enhance the accuracy of sex estimation in the Korean population and improve its applicability in practical forensic contexts.

This study aimed to analyze the 3D morphological characteristics of the condylar part of the occiput in a Korean population and to evaluate the accuracy of sex estimation by deriving discriminant functions based on distance and area measurements. Through this approach, the forensic anthropological applicability of skull base structures, which have been relatively underutilized, was assessed, and a methodological basis for sex estimation applicable to partially preserved remains was established.

Material and Methods

1. Participants

This study was conducted as a 3D analysis using cranial computed tomography (CT) images from 87 Korean individuals (44 males and 43 females) of known age and sex. The use of imaging data for research purposes was approved, and images were collected between 2008 and 2009 with the cooperation of the Department of Neuroradiology at Seoul St. Mary's Hospital. Only patients without cranial trauma or deformities were included in the analysis. The mean age of the subjects was 42.4 years (males: 43 years; range, 20-67 years; females: 41.7 years; range, 20-66 years). All procedures were performed in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (IRB) of the Catholic University of Korea (approval no. MC24EASI0004). The requirement for informed consent was waived by the IRB.

2. 3D reconstruction and measurements

CT examinations for forensic evaluation were performed using a scanner (Siemens Healthcare, Erlangen, Germany). Imaging parameters were set as follows: tube voltage 120 kVp, tube current 210 mAs, slice thickness 0.75 mm, pitch factor 0.35, increment 0.7 mm, and rotation time 0.3 seconds. The total scan time ranged from 60 to 75 seconds, depending on the body height. Subsequently, the DICOM data obtained from CT were processed using the 3D image-based engineering software MIMICS version 23.0 (Materialise NV, Leuven, Belgium) with a threshold range of 673-3,071 Hounsfield units (HU). The cranial images were segmented and converted into a computer-aided format for further analysis. All measurements and their definitions are presented in Table 1 and Figs. 1, 2.

Fig. 1.

Distance measurements in the condylar part of the occiput. A basal view of the reconstructed occipital condyles and foramen magnum showing the landmarks and distance variables. The 3D models of the skull base were generated using MIMICS (Materialise Mimics version 23.0, Materialize NV, Leuven, Belgium), a 3D image-based engineering software. AOC, distance between most anterior points of the occipital condyle; FML, maximum length of the foramen magnum; FMW, maximum width of the foramen magnum; MBB, maximum bicondylar breadth; OCL, maximum length of the occipital condyle; OCW, maximum width of the occipital condyle; POC, distance between most posterior points of the occipital condyle.

Fig. 2.

Area measurements in the condylar part of the occiput (orange). (A) Articular surface area of the occipital condyles. (B) Foramen magnum area.

Table 1.

List of morphometric parameters measured

Abbreviation	Morphometric parameters
MBB	Maximum bicondylar breadth
AOC	Distance between most anterior points of the occipital condyles
POC	Distance between most posterior points of the occipital condyles
OCL	Maximum length of the occipital condyle
OCW	Maximum width of the occipital condyle
OCA	Articular surface area of the occipital condyle
FML	Maximum length of the foramen magnum
FMW	Maximum width of the foramen magnum
FMA	Area of the foramen magnum

Listed in anatomical order, not alphabetical order.

3. Statistical analysis

All statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Intraobserver reliability was assessed by a single observer (S.J. Choi) with advanced training and experience in 3D measurements, who evaluated all variables across all samples twice. The intraclass correlation coefficient (ICC) was examined using a two-way random-effects model with absolute agreement for single measurements (ICC (2,1)). ICC values were interpreted as follows: <0.50, poor; 0.50-0.75, moderate; 0.75-0.90, good; and >0.90, excellent. Descriptive statistics, including means, standard errors, and Cohen's d to quantify effect sizes, were calculated for 12 variables derived from the length and width of the occipital condyle and foramen magnum, as well as for the area of each structure. Differences in mean values between the sexes were analyzed using independent t-tests.

Additionally, normality, skewness, and kurtosis were evaluated to ensure the reliability of the discriminant function analysis. Given the number of measurement variables, normality was assessed using the Shapiro-Wilk test. Statistical significance was set at P<0.05. Discriminant analysis was performed using univariate and multivariate approaches, with variables analyzed separately for the left and right sides, including only those that met the acceptable skewness and kurtosis criteria and were statistically significant (P<0.05).

Results

The results of the intraobserver reliability analysis are presented in Table 2. The ICC (2,1) values obtained by the same observer exceeded 0.90, indicating excellent reliability according to the criteria proposed by Koo and Li [14]. This suggests minimal measurement variability and high intra-observer reproducibility.

Table 2.

Assessment of intra-observer reliability for condylar part of the occiput measurements (n=87)

Measurement	ICC	95% CI
AOC	0.989	0.984-0.993
FMA	0.998	0.997-0.999
FML	0.992	0.983-0.995
FMW	0.984	0.976-0.990
MBB	0.994	0.989-0.997
OCA (Lt)	0.990	0.982-0.994
OCA (Rt)	0.991	0.984-0.995
OCL (Lt)	0.991	0.985-0.994
OCL (Rt)	0.990	0.984-0.994
OCW (Lt)	0.984	0.976-0.990
OCW (Rt)	0.986	0.979-0.991
POC	0.984	0.976-0.990

ICC, intraclass correlation coefficient; CI, confidence interval; AOC, distance between most anterior points of the occipital condyle; FMA, area of the foramen magnum; FML, maximum length of the foramen magnum; FMW, maximum width of the foramen magnum; MBB, maximum bicondylar breadth; OCA, articular surface area of the occipital condyle; OCL, maximum length of the occipital condyle; OCW, maximum width of the occipital condyle; POC, distance between most posterior points of the occipital condyle.

The distribution of the data was assessed using normality, skewness, and kurtosis tests, as presented in Table 3. Given the sample size within each group (n < 50), the Shapiro-Wilk test was used to evaluate normality. Most variables satisfied the assumption of normal distribution, except for maximum width of the foramen magnum (FMW), area of the foramen magnum (FMA), articular surface area of the occipital condyle (OCA) (Lt), and OCA (Rt) in the female group. Although the Shapiro-Wilk test indicated deviations from normality for these variables, their skewness and kurtosis values were within acceptable ranges, suggesting approximate normality.

Table 3.

Skewness, kurtosis, and normality test in condylar part of the occiput measurements (n=87)

Variable	Sex	Skewness	Kurtosis	Shapiro-Wilk
AOC	Male	-0.295	0.340	0.771
	Female	0.120	-0.558	0.603
FMA	Male	-0.030	-0.507	0.538
	Female	1.592	5.499	<0.001^a)
FML	Male	-0.328	-0.412	0.600
	Female	0.657	1.573	0.182
FMW	Male	-0.164	-0.307	0.575
	Female	1.206	3.069	0.002^a)
MBB	Male	0.308	0.763	0.747
	Female	-0.047	0.573	0.705
OCA (Lt)	Male	0.239	-0.750	0.485
	Female	-0.023	-1.348	0.015^a)
OCA (Rt)	Male	0.625	0.659	0.320
	Female	0.812	0.231	0.017^a)
OCL (Lt)	Male	0.253	-0.226	0.724
	Female	0.183	-0.134	0.747
OCL (Rt)	Male	0.453	-0.072	0.575
	Female	0.018	-0.238	0.991
OCW (Lt)	Male	0.293	0.123	0.918
	Female	0.323	-0.095	0.348
OCW (Rt)	Male	0.086	-0.743	0.455
	Female	0.045	-0.443	0.906
POC	Male	0.329	-0.069	0.516
	Female	-0.631	0.379	0.240

AOC, distance between most anterior points of the occipital condyle; FMA, area of the foramen magnum; FML, maximum length of the foramen magnum; FMW, maximum width of the foramen magnum; MBB, maximum bicondylar breadth; OCA, articular surface area of the occipital condyle; OCL, maximum length of the occipital condyle; OCW, maximum width of the occipital condyle; POC, distance between most posterior points of the occipital condyle.

^a) P<0.05.

Most measurements of the condylar part of the occiput exhibited sex-related differences in the Korean sample, with males generally showing larger values than females (Table 4). This pattern was consistently observed in the descriptive statistics and independent t-test results, with most variables reaching statistical significance (P<0.05) except for distance between most anterior points of the occipital condyle (AOC). Additionally, paired t-tests were conducted to assess side differences in the occipital condyles, revealing significant left-right asymmetry in most variables, with the exception of OCA.

Table 4.

Descriptive statistics of the variables in the condylar part of the occiput (n=87)

Variable	Sex	Mean	SD	SE	Cohen's d	t-value	P-value
AOC	Male	21.109	2.706	0.408	0.256	-1.191	0.237^b)
	Female	21.698	1.808	0.276
FMA	Male	810.972	111.022	16.737	0.608	2.831	0.006
	Female	747.208	98.524	15.025
FML	Male	35.606	3.064	0.462	0.520	2.424	0.017
	Female	34.095	2.737	0.417
FMW	Male	30.101	2.380	0.359	0.464	2.165	0.033
	Female	29.090	1.948	0.297
MBB	Male	48.161	2.710	0.409	1.569	7.308	<0.001
	Female	44.079	2.491	0.380
OCA (Lt)	Male	293.942	35.578	5.364	2.650	12.357	<0.001
	Female	204.659	31.651	4.827
OCA (Rt)	Male	301.346	38.990	5.878	2.723	12.683	<0.001
	Female	207.332	29.356	4.477
OCL (Lt)	Male^a)	20.843	1.949	0.294	1.584	7.398	<0.001
	Female^a)	17.522	2.232	0.340
OCL (Rt)	Male^a)	20.753	1.820	0.274	1.671	7.791	<0.001
	Female^a)	17.396	2.186	0.333
OCW (Lt)	Male^a)	14.081	1.491	0.225	1.575	7.343	<0.001
	Female^a)	11.788	1.418	0.216
OCW (Rt)	Male^a)	14.103	1.508	0.227	1.474	6.872	<0.001
	Female^a)	12.009	1.327	0.202
POC	Male	42.432	3.202	0.483	0.665	3.101	0.003
	Female	40.637	2.063	0.315

^a) P<0.05 in the paired t-test between the right and left sides;

^b) Not statistically significant (P>0.05).

The discriminative performance of each variable was evaluated using eigenvalues and Wilks’ lambda, with a demarcation point of 0, where values greater than 0 were classified as male, and values less than 0 as female (Table 5). Variables showing larger eigenvalues and smaller Wilks’ lambda values, indicating greater discriminative power between the sexes, included maximum length of the occipital condyle (OCL) (Lt), OCL (Rt), maximum width of the occipital condyle (OCW) (Lt), OCW (Rt), maximum bicondylar breadth (MBB), OCA (Lt), and OCA (Rt), all of which demonstrated relatively higher classification accuracy than the other variables. In contrast, all variables related to the foramen magnum exhibited very low classification accuracy, regardless of whether distance measurements (65.5% for maximum length of the foramen magnum and 62.1% for FMW) or area measurements (66.7% for FMA) were used.

Table 5.

Canonical discriminant function coefficients, classification accuracy, and cross-validation of the condylar part of the occiput measurements (n=87)

Variable	Eigen value	Wilks's lambda	Canonical correlation	Discriminant function		Demarcation point	Cross-validated accuracy (%)
Variable	Eigen value	Wilks's lambda	Canonical correlation	Statistics	Constant	Demarcation point	Male	Female	Overall
AOC	0.017	0.984	0.128	0.434	-9.277	F < 0 < M	54.5	53.5	54.0
FMA	0.094	0.914	0.294	0.010	-7.421	F < 0 < M	63.6	69.8	66.7
FML	0.069	0.935	0.254	0.344	-11.990	F < 0 < M	63.6	67.4	65.5
FMW	0.055	0.948	0.229	0.459	-13.597	F < 0 < M	56.8	67.4	62.1
MBB	0.628	0.614	0.621	0.384	-17.718	F < 0 < M	79.5	83.7	81.6
OCA (Lt)	1.796	0.358	0.801	0.030	-7.414	F < 0 < M	90.9	95.3	93.1
OCA (Rt)	1.893	0.346	0.809	0.029	-7.373	F < 0 < M	86.4	90.7	88.5
OCL (Lt)	0.644	0.608	0.626	0.478	-9.173	F < 0 < M	81.8	74.4	78.2
OCL (Rt)	0.714	0.583	0.645	0.498	-9.504	F < 0 < M	81.8	79.1	80.5
OCW (Lt)	0.634	0.612	0.623	0.687	-8.895	F < 0 < M	77.3	83.7	80.5
OCW (Rt)	0.556	0.643	0.598	0.704	-9.195	F < 0 < M	79.5	76.7	78.2
POC	0.113	0.898	0.319	0.370	-15.387	F < 0 < M	61.4	62.8	62.1

Multivariate discriminant analysis was performed using the enter method (Table 6) with models constructed by including and excluding area measurements. Overall, the multivariate models demonstrated improved classification accuracy compared to the univariate analyses, while maintaining comparable performance across models with different variable compositions.

Table 6.

Multiple discriminant function analysis using the enter method for the condylar part of the occiput, with and without inclusion of area measurements (n=87)

Functions	Variables	Discriminant function	Group centroid		Cross-validated accuracy (%)
Functions	Variables	Discriminant function	Male	Female	Male	Female	Overall
With area measurements	AOC	0.032	1.825	-1.868	100.0	95.3	97.7
	FML	0.064
	FMW	-0.106
	MBB	0.007
	OCL (Lt)	-0.034
	OCL (Rt)	0.175
	OCW (Lt)	0.159
	OCW (Rt)	0.296
	POC	-0.071
	FMA	0.003
	OCA (Lt)	0.008
	OCA (Rt)	0.016
	Constant	-14.357
Without area measurements	AOC	0.068	1.390	-1.423	97.7	93.0	95.4
	FML	0.103
	FMW	-0.008
	MBB	0.042
	OCL (Lt)	0.055
	OCL (Rt)	0.360
	OCW (Lt)	0.199
	OCW (Rt)	0.460
	POC	-0.155
	Constant	-16.837

In addition, to account for fragmented conditions, separate multivariate discriminant analyses were performed for each component of the condylar part of the occiput (Table 7). Using the enter method for each component, the classification accuracy for the left and right occipital condyles generally improved compared with the univariate analysis, although the multivariate model for the left occipital condyle showed a slight decrease in male classification accuracy despite improved female classification accuracy, whereas that for the foramen magnum showed no change, even in the multivariate model.

Table 7.

Multiple discriminant function analysis using the enter method for the three components of the condylar part of the occiput, including the left and right occipital condyles and foramen magnum (n=87)

Components	Variables	Discriminant function	Group centroid		Cross-validated accuracy (%)
Components	Variables	Discriminant function	Male	Female	Male	Female	Overall
Left occipital condyle	OCL (Lt)	0.121	1.441	-1.475	86.4	97.7	92.0
	OCW (Lt)	0.245
	OCA (Lt)	0.022
	Constant	-10.952
Right occipital condyle	OCL (Rt)	0.207	1.607	-1.645	93.2	93.0	93.1
	OCW (Rt)	0.341
	OCA (Rt)	0.020
	Constant	-13.411
Foramen magnum	FML	0.010	0.301	-0.308	61.4	72.1	66.7
	FMW	-0.043
	FMA	0.010
	Constant	-6.845

OCL, maximum length of the occipital condyle; OCW, maximum width of the occipital condyle; OCA, articular surface area of the occipital condyle; FML, maximum length of the foramen magnum; FMW, maximum width of the foramen magnum; FMA, area of the foramen magnum.

Discussion

Sex determination is a fundamental step in forensic anthropology [15]. Although the skull and pelvis are the most reliable indicators of sexual dimorphism [16], they are not always preserved intact, necessitating the use of alternative anatomical regions under fragmented conditions [17,18]. This study focused on the condylar part of the occiput, which is relatively well protected and may remain preserved even in fragmented remains [19]. The results demonstrated measurable sex-related differences in the Korean population in this region and confirmed its potential utility for sex estimation. Given the limited use of area-based parameters in previous studies, these findings highlight the value of incorporating such measurements and supporting the use of the condylar part of the occiput, particularly the occipital condyles, as a practical alternative when conventional skeletal markers are not available.

In this study, independent t-tests between the sexes revealed significant morphological differences in the condylar part of the occiput (Table 4). In particular, significant sexual dimorphism was observed in the occipital condyles, with distance measurements (OCL and OCW) and area measurements (OCA) showing statistically significant differences between the sexes. The foramen magnum also showed statistically significant sex-related differences, although with lower t-values and higher p-values than the occipital condyle, suggesting a limited contribution to sex estimation. Notably, the finding that the occipital condyle shows greater sexual dimorphism than the foramen magnum is consistent with the results reported across diverse populations [20-22].

Analysis of bilateral symmetry in the occipital condyles within the condylar part of the occiput showed left-right asymmetry in distance measurements, whereas area measurements remained symmetric (Table 4). This finding suggests that although the length and width of the occipital condyle differ between sides, the surface area remains comparable, indicating that these differences are attributable to shape variations that are not fully reflected by simple distance measurements alone. Although no previous studies assessing area measurements have been identified, the observed left-right asymmetry of the occipital condyle based on distance measurements is consistent with previous reports that attributed condylar asymmetry to habitual posture, muscle use patterns, and functional laterality [23-25].

In this study, the univariate discriminant analysis demonstrated that variables related to the length, width, and area of the occipital condyle exhibited greater discriminative power than those associated with the foramen magnum (Tables 4, 5). In particular, the AOC did not show a significant sex difference in the independent t-test (P=0.237) and yielded the lowest classification accuracy (54.0%). This low classification performance may be attributed to its limited discriminative power, as indicated by the small eigenvalue (0.017) and relatively high Wilks’ lambda (0.984). In contrast, MBB and distance between most posterior points of the occipital condyle (POC) exhibited significant differences and achieved higher classification accuracies (81.6% and 62.1%, respectively). Notably, MBB, which represents the overall width of the condylar part, exhibited a larger effect size (Cohen's d=1.569) than AOC (0.256) and POC (0.665), indicating less overlap between the groups. These findings are consistent with previous morphometric evidence, suggesting that sexual dimorphism of the occipital condyle involves differences in size and shape [26]. Variables related to the foramen magnum showed statistically significant differences in the independent t-test; however, their univariate classification accuracy was relatively low. This may be attributed to the substantial overlap between the groups and the relatively small effect sizes. Additionally, the influence of the foramen magnum shape should be considered given the difficulty of representing egg-shaped, hexagonal, and elliptical configurations using simple linear measurements alone [10]. These findings indicate that, in addition to data conditioning procedures, such as reducing sample heterogeneity by restricting the age range, excluding pathological or deformed cases, controlling for body size or body mass index, and ensuring population homogeneity, nonmetric morphological assessments should be incorporated alongside metric approaches to overcome the limitations of linear measurements in anthropological analysis.

In this study, multivariate discriminant functions were developed to reflect the practical forensic conditions (Table 6). Separate models were constructed depending on whether area measurements could be obtained from CT imaging, or only distance measurements were available from dry bone, acknowledging the variability in laboratory conditions. Furthermore, additional multivariate analyses were performed for each component of the condylar part of the occiput to account for fragmented remains (Table 7), enabling its application even when only partial structures were recovered. These results emphasize that in forensic contexts where preservation is limited, analytical approaches should be flexibly adapted to the available remains. In such situations, the reliance on a single skeletal element may be insufficient, and a multifactorial approach incorporating multiple anatomical regions and measurements may improve the accuracy and reliability of sex estimation, as emphasized in age estimation in forensic anthropology [27].

In conclusion, the 3D morphological characteristics of the condylar part of the occiput were analyzed in a Korean population, and discriminant functions for sex estimation were derived. Sexual dimorphism was more pronounced in the occipital condyles than in the foramen magnum, indicating that the occipital condyles are more informative for sex estimation. In contrast, the foramen magnum showed limited discriminative value, suggesting the need for further investigation, particularly regarding sex-related differences in intracranial cavity volume. Future studies incorporating biomechanical approaches from a functional perspective in addition to morphometric analyses are expected to provide a deeper understanding of sexual dimorphism in the condylar part of the occiput.