https://orcid.org/0000-0002-0355-1006
Figures
Abstract
Background
Sex estimation is an essential topic in the field of individual identification in forensic anthropology. Recent studies have investigated a growing range of techniques for estimating sex from human skulls.
Objectives
This study aims to provide a scoping review of the literature on techniques used in skull-based sex estimation, serving as a valuable reference for researchers.
Sources of evidence
The literature search was performed using PubMed, Scopus, and Web of Science from January 2020 to February 2024.
Eligibility criteria
Eligible studies have investigated issues of interest to forensic anthropology about sex estimation using skull samples.
Charting methods
A total of 73 studies met the inclusion criteria and were categorized and analyzed based on the anatomic sites, modalities, trait types, and models. Their accuracy in estimating sex was subsequently examined, and the results were charted.
Results and conclusions
Our review highlights that the 3D medical imaging technique has enhanced the efficiency and stability of skull-based sex estimation. It is anticipated that advancements in 3D imaging and computer vision techniques will facilitate further breakthroughs in this field of research.
Citation: Wang X, Liu G, Wu Q, Zheng Y, Song F, Li Y (2024) Sex estimation techniques based on skulls in forensic anthropology: A scoping review. PLoS ONE 19(12): e0311762. https://doi.org/10.1371/journal.pone.0311762
Editor: Rijen Shrestha, The Maldives National University, NEPAL
Received: August 6, 2024; Accepted: September 24, 2024; Published: December 9, 2024
Copyright: © 2024 Wang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study was supported by the National Natural Science Foundation of China (No. 82202079) and Natural Science Foundation of Sichuan Province (No. 2022NSFSC1403), both awarded to Ms Yuan Li. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Forensic anthropology often involves estimating sex, ancestry, age, and height from the skeletal remains of unknown individuals. This estimated summary is referred to as the biological profile, which is subsequently used to compare with missing person records to achieve personal identification [1]. Successfully estimating sex is a prerequisite for developing a reliable biological profile because the estimation of stature, age, and ancestry follows patterns related to sex [2].
With regard to human skeletons, morphological variations have been attributed to diverse factors. These include but are not limited to, the unique developmental pathways specific to each sex, hormonal responses to environment stimulation, and the adaptability shaped by sexual labour division in past societies [3–5]. These factors contribute to the inherent variations in the appearance between male and female skeletons, thereby enabling skeletal structures to be informative indicators in sex estimation [6]. Numerous published studies investigate various skeletal features and their capacity to differentiate between sexes effectively. These features encompass, but are not limited to, the pelvis [7–9], skull [2, 4, 10], ribs [11, 12], sternum [13–15], vertebrae [16–18], clavicle [19–21], and limb bones [22–24]. Among these, the pelvis is referred to as the most reliable bone for sex estimation [7, 25, 26]. Nevertheless, the skull is considered the most reliable alternative structure when the pelvis is unavailable for analysis or displays ambiguous signals [27].
The skull, comprising both the cranium and the mandible, presents an interesting set of traits for studying morphological variation in relation to the sexes or other factors, such as genetics, developmental processes, dietary habits, and environmental influences [4, 13]. On average, the male skull is larger and thicker, whereas the female skull is smaller and smoother [4, 28, 29]. Sexual dimorphism in cranial morphology is evident not only in relative size but also in distinct shape features. For instance, males tend to develop a more prognathic (with a protruding jaw) and dolichocephalic (elongated head shape) cranium compared to females [30, 31]. As per recent research, the shape of the skull, particularly the cranial and facial bones, is more likely to provide potentially useful information for sex estimation [4].
Sex estimation based on skulls involves three fundamental factors: modalities, methods, and analysis models. Fig 1 illustrates the process of sex estimation for better understanding.
The orange box illustrates the skull modalities available for sex estimation, including dry skulls, radiographs, Computed Tomography (CT), and Magnetic Resonance Imaging (MRI) scans, as well as three-dimensional (3D) models derived from CT reconstruction or 3D scanning. The blue box outlines non-metric, traditional metric, and geometric morphometric (GM) methods for sex estimation. Within GM methods, there are further subdivisions into landmark-, outline-, and surface-based approaches. The green box shows analytical models utilized for sex estimation. Statistical analysis includes discriminant function analysis and logistic regression algorithms, while classical machine learning involves multiple algorithms, with support vector machine, neural network, decision tree, and random forest presented in the diagram as representative examples. Deep learning, featured in both blue and green boxes, stands out for its ability to automatically extract features and classify sexes. By combining different modalities, methods, and models, a broad range of sex estimation outcomes can be achieved.
Modalities
Forensic anthropologists have traditionally preferred to obtain primary data from skull collections. The drawback of metric methods based on skull collections is that the manipulation may cause secondary damage to the bone. Fortunately, the three-dimensional (3D) scanning technique addresses this issue by providing digital 3D models of dry skulls.
Digital radiography provides architectural and morphological details of the skull and thus reveals multiple anatomical landmarks for comparison, making it ideal for sex estimation in charred or decomposed cases [32, 33]. However, radiographs present the 3D characteristics of the skull on a two-dimensional (2D) image. Thus, the morphometric traits of the skull superstructures and intracranial structures sometimes appear ambiguous on a 2D roentgenogram [34].
The Computed Tomography (CT) images from multi-planar reformation and Magnetic Resonance Imaging (MRI) scans offer high-resolution views of the targeted plane and the internal edges of the skull, thereby facilitating precise positioning of landmarks and measurements [35]. However, these sectional images still represent the cranial morphology at a 2D level, posing difficulties in accurately measuring the cranial surface. The CT reconstruction using the volume rendering technique offers a significant opportunity to overcome the limitations of 2D imaging by modeling skull collections at the 3D level. These 3D models provide a more comprehensive and detailed understanding of cranial morphology [35–38]. With the aid of computer visual techniques, 3D models hold the potential to reveal more cranial traits, further enhancing the accuracy of sex estimation.
Non-metric and metric methods
Sex estimation methods are generally classified into non-metric (morphologic) and metric [3]. Morphological methods rely on visual assessment of the various structures and configurations of skulls. Walker’s method [39], which relies on standard morphological traits originally described by Buikstra and Ubelaker [40], is one of the most widely used morphological methods for estimating sex from the skull. This standard comprises five traits: nuchal crest prominence, mastoid process size, supraorbital edge thickness, supraorbital ridge/glabella prominence, and mental eminence. Each variable is scored ordinally on a scale ranging from 1 (most gracile) to 5 (most robust). Walker [39] developed this standard using samples from North America and Britain, achieving an accuracy rate of 89%. These non-metric traits allow for rapid preliminary sex estimation results without using any special equipment; however, they are inherently observer-dependent and necessitate substantial scoring experience for cranial features [36, 41, 42].
Metric evaluation is founded on the quantification of sexual dimorphism in skull dimensions, yielding more objective results for sex estimation [43, 44]. Nevertheless, it often necessitates precise localization of anatomical landmarks. Metric methods could be further divided into traditional metrics and geometric morphometrics (GM). Traditional metric methods develop statistical models based on cranial indicators derived from manual or computer-aided measurement techniques, such as linear distances, angles, areas, and indices.
GM enables the morphology quantification of rigid structures of skulls and the comparison of the size and shape separately between males and females. The anatomical landmarks, outlines, and homologous surfaces are employed to capture the geometric differences in sexually dimorphic structures. Landmark-based geometric morphometric approaches typically employ landmarks as Cartesian coordinates to quantify the shape difference of skulls. Generalized Procrustes analysis (GPA) is most commonly used to conduct the landmark configuration superimposition, in order to eliminate the differences in the location, scale, and size across configurations. After landmark standardization, the consensus mean configurations for males and females will be generated. Following that, shape variation among superimposed configurations between the target skull and the mean male/female shape can be quantified as Procrustes distances [3, 45]. These shape variation parameters provide objective indicators to estimate the sex of the target skull.
Compared with landmarks, outline-based methods apply landmarks to represent contours or boundary outlines. They are independent of the presence of true anatomical landmarks, thus providing a more complex, quasicontinuous representation of sex variation. As a curve-fitting function, Fourier analysis decomposes spatial information about an outline into an infinite series of sine and cosine functions, weighted by Fourier descriptors that represent shape variables [46].
The surface-based approach involves employing homologous models with numerous vertices to represent or approximate the surface of 3D digital skulls. Surface registration techniques such as the iterative closest point and coherent point drift [47] are employed to mitigate deviations caused by unequal location and rotation. The enormous amount of vertex data on the 3D skull surface is reduced to lower dimensions by principal component analysis before statistical processing. Although it doesn’t rely on landmarking, analyzing massive point data demands robust predictive models and increased computational power.
Analytical models
The analytical models for sex estimation can be roughly divided into statistical learning, classical machine learning (ML), and deep learning (DL), with the latter two collectively referred to as artificial intelligence (AI).
Statistical learning includes discriminant function analysis (DFA) and logistic regression (LR), which is the simple choice for sex estimation. DFA projects m-dimensional data into a space where the results have the properties of maximum discrimination between classes and minimal intraclass distance [48]. A linear function is formulated by combining a set of variables to calculate the classification score for each sample. This score is subsequently compared with the threshold determined from the group centroids [49]. LR predicts the probability of an event occurring by fitting a sigmoid function to the data and has been regularly employed in metric indicators for sex estimation [48].
Classical ML, such as the support vector machine (SVM), decision trees (DT), and neural networks (NN), have been used for skull-based sex estimation. Unlike statistical techniques that operate solely on group parameters, such as means and covariance matrices, classical ML algorithms are capable of creating nonlinear decision boundaries between groups with minimal knowledge of the domain or presumptions about the probability distributions underlying the observed data [50].
DL, composed of multiple neural network layers, allows for extraction and learning representations of training data with multiple levels of abstraction [51]. As one type of DL algorithm, the convolutional neural network (CNN) is designed to learn spatial hierarchies of features from visual data, such as images. The convolution layer is a fundamental component of CNN architecture that applies convolution kernels to extract features from each position on images [52]. Compared with statistical and classical ML learning, CNN can perform feature extraction automatically without hand-crafted involvement, due to the utilization of convolution layers. It has proven itself a powerful method in medical fields that rely on imaging data, including radiology, pathology, dermatology, and ophthalmology [22, 53–56]. However, CNN models are currently seldom employed in the field of sex estimation using skulls, which may be partly explained by the necessity for extensive training datasets and a profound level of technical expertise in AI.
Objectives
The selection of modalities, methods, and analytical models for skull-based sex estimation is diverse. Consequently, studies in this field are vast and intricate, yielding a broad spectrum of outcomes. Our review aims to summarize the various modalities, non-metric and metric methods, predictive models, and software employed by recent studies, highlighting innovative techniques to enhance the reliability and accuracy of sex estimation.
Materials and methods
Protocol and registration
This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-analysis Protocols extension for scoping reviews (PRISMA-ScR) [57]. The checklist of the PRISMA-ScR guidelines is provided in the S1 Table.
Search process
The search was conducted using PubMed, Scopus, and Web of Science databases. The search string included keywords related to the study’s aim: (’sex’ OR ’gender’) AND (’determination’ OR ’estimation’ OR ’prediction’) AND ’skull’. Filters were applied to limit the search results to articles published between January 2020 and February 2024 in English and Chinese.
Eligibility criteria
The following studies were included: (a) original articles; (b) studies published in Chinese or English; (c) studies using craniums, mandibles, or whole skulls for sex estimation. The exclusion criteria included the following: (a) non-human subjects; (b) conference papers, reviews, case reports, and books; (c) studies using other bones (including teeth) other than skulls; (d) intact or fractured skulls; (e) studies using samples of sub-adults under the age of 18 years old; (f) studies without disclosing the outcomes of their sex prediction models.
Study selection
After the search was completed, duplicates were removed, and relevant articles were selected from the titles and abstracts. The inclusion and exclusion criteria listed for this systematic review were considered while reading the full text to ascertain eligibility. Two independent researchers (XW and GL) carried out the systematic search, study selection, and data extraction. Disagreements among researchers were resolved by consensus, mediated by a third reviewer (YZ).
Results
Selection of sources of evidence
Our preliminary search yielded 2200 records in total. Following the duplicates’ elimination and automatic filtering, 716 studies remained. After reviewing the titles and abstracts, we excluded 507 articles from further consideration. A total of 136 full-text articles were excluded since they failed to meet the inclusion or exclusion criteria, resulting in a final inclusion of 73 studies in this scoping review. The procedure of the literature search is summarized in Fig 2. The specific characteristics of each study are summarized in the S2 Table.
Synthesis of results
Populations.
The majority of studies on sex estimation were conducted on Asian populations, with India [58–71] leading the way, followed by Turkey [35, 72–78], Iran [79–84], and China [85–87]. In Europe, notable contributions came from Bosnia [88–92] and Bulgaria [36, 93–96], while a single study encompassed four distinct Western European populations: Belgium, Switzerland, France, and Portugal [97]. Several studies were conducted in African countries, including Nigeria [98–100] and Egypt [101, 102], and in American countries, including Brazil [103, 104] and the United States [105]. Notably, one study concerned Caucasians, but the countries where this study was conducted were not specified [37]. Fig 3 shows the detailed distribution of the studied population.
Fig 3 depicts the geographical distribution of the populations investigated across 73 studies, where darker shades of blue signify a greater frequency of research conducted within the respective region.
Modalities.
Among the included studies, CT was the most frequently employed imaging modality, particularly in the form of 3D skull models derived from CT reconstructions (Fig 4A). Moreover, one study conducted by Kumar et al. utilized both 2D CT scans and 3D CT models of 217 Indian skulls for metric analysis, achieving an accuracy rate of 93.6% [63]. In contrast, MRI technology was solely employed in the study by Liu et al. [85], where they assessed sexual dimorphism in the sphenoid sinus of Chinese individuals, albeit with a relatively low accuracy of 63.3%.
(a) Numbers of included studies using different modalities for sex estimation; (b) Sample sizes of studies using different modalities; (c) Accuracy corresponding to each type of modality.
The sample sizes in these studies varied significantly, ranging from 34 to 1234 cases. Notably, the sample sizes for CT data tended to be larger than those for dry skulls and radiographs (Fig 4B). The study with the largest sample size came from Kondou et al. [106], who utilized CNN algorithm to conduct sex estimation on 3D CT skull data from 1234 Japanese individuals. This resulted in exceptionally high accuracy rates of 95.0% and 93.0% for the training and testing datasets, respectively.
The accuracy of 3D models derived from CT reconstruction was consistently high and outperformed that of other modalities (Fig 4C). Despite the fact that five studies focusing on sex estimation in partial skull regions reported accuracies falling below 75% [74, 107–110], the majority of studies that employed 3D CT models consistently achieved accuracies exceeding 85%. In terms of dimensionality, both CT slices and radiographs, being 2D planar images, exhibit reduced accuracy in sex estimation compared to 3D models.
Non-metric and metric methods.
The majority of studies utilized traditional metric traits, followed by GM and non-metric methods. Additionally, one study by Jeong et al. [107] employed both traditional metric and non-metric methods for sex estimation (Fig 5A). Notably, the study by Koudou et al. [106] based on CNNs was excluded, as the network automatically extracted the cranial features, falling outside the categories of metric or non-metric traits.
(a) Numbers of studies using different methods for sex estimation; (b) Accuracy distribution corresponding to each method on 3D cranial models; (c) Anatomical regions used by traditional metric studies. The rings are labelled with the names of anatomical sites and the accuracy achieved in studies.
To decrease the number of variables, we focused our comparison on studies that employed 3D skulls as the modalities, including dry skulls, digital models from 3D scanning, and 3D CT models. Among the studies we reviewed, all non-metric [76, 87, 104, 111–115] and GM studies [36, 84, 86, 88–92, 94–96, 109, 110, 116, 117] and 19 out of the 47 metric studies employed 3D cranial models. Additionally, the only study incorporating metric and non-metric approaches also utilized 3D models for its analysis [107]. All three methods exhibited a broad range of accuracy levels, yet their accuracy rates’ differences appeared ambiguous (Fig 5B).
Among traditional metric studies, we identified nine partial skull regions that were investigated for sexual dimorphism, underscoring their significance in sex estimation (Fig 5C). Notably, the foramen magnum (FM) and occipital condyles (OC) emerged as the most frequently examined anatomical landmarks, closely followed by the mandibles and mastoid processes. Nevertheless, despite their prevalence, neither FM nor OC achieved a high level of accuracy, failing to exceed 80% in distinguishing between males and females. In contrast, Farhadian et al. [79] employed measurements derived from the mastoid process, reporting the highest accuracy rate of 97.5% for sex estimation.
Analytical models.
Seventy-six studies employed statistical learning approaches, while seven solely utilized AI algorithms [36, 86, 101, 106, 110, 116, 118]. Moreover, ten articles incorporated both ML and statistical learning methods (Fig 6A) [75, 79, 83, 93, 96, 103, 111, 115, 117, 119]. This section specifically examined the accuracy of the models employed in these ten studies. Interestingly, when compared to simple statistical learning models, the accuracy of sophisticated machine learning models did not increase (Fig 6B). Furthermore, it is not feasible to assess DL’s performance in this situation, given that only one study employed the DL algorithm [106].
(a) Numbers of included studies using different predictive models for sex estimation. AI: Artificial Intelligence; (b) Accuracy corresponding to each type of analytical model. ML: Machine Learning.
Software for skull-based sex estimation
From the included studies, we identified three software tools for sex estimation using skulls. The CroWalker software was developed by Bareša et al. [114] utilizing 200 MSCT scans from the modern Croatian population. This software employs four non-metric traits- nuchal crest, mastoid process, supraorbital edge, and glabella- to estimate sex based on LR equations. The scoring of these traits follows the guidelines in the MorphoPASSE user manual. CroWalker allows users to determine the probability of an individual being male or female, leading to a final sex classification. In Bareša et al.’s study [114], the best-performing equation, which included the glabella and mastoid process, achieved an accuracy rate of 86.25%.
The SexEst software, developed by Constantinou and Nikita [120], is accessible online at http://sexest.cyi.ac.cy/. This software utilizes 43 variables to measure cranial and postcranial bones, drawing upon the Goldman dataset (Auerbach & Ruff, 2004) and the Howells dataset (Howells, 1973; Howells, 1989; Howells, 1995). Additionally, SexEst integrates various classification models, including DFA, SVM, random forest (RF), and gradient boosting (GB), allowing users to choose the model that best suits their data. Nikita, P.A. et al. [119] conducted a study to assess the efficacy of SexEst on Greek individuals, and the light GB model achieved the highest accuracy rate of 73.5% for sex classification.
KKU Sex Estimation, developed by Techataweewan et al. [43], is available at https://tinyurl.com/26t3x3mu. This software is grounded on the dataset consisting of 322 Thai skull collections sourced from Khon Kaen University. Using the DFA model, it can handle 25 cranial and 5 mandibular measurements or only a single measurement. The output includes the predicted sex, measures of model success, and a histogram showing the position of the “Unknown” within the distribution of the reference samples. In the study by Techataweewan et al. [43], the accuracy rate of sex estimation was 92.1%.
Discussion
A substantial number of studies have addressed the issue of sex estimation based on the features of human skulls between January 2020 and February 2024, demonstrating the significance of this topic in forensic anthropology. The development in medical imaging technology has significantly broadened the scope of modalities available for sex estimation research, such as X-rays, CT, and MRI scans. Both non-metric and metric approaches coexist and complement each other, revealing the sexual dimorphism in cranial morphology. Furthermore, AI algorithms offer more advanced and sophisticated model architectures for sex estimation, capable of analyzing the intricate data acquired through medical imaging technologies. This study overviewed the recent advancements and outcomes of skull-based sex estimation technology. Our findings underscore the effectiveness of 3D cranial models in enhancing the accuracy of sex estimation and emphasize the need to further explore the untapped potential of AI algorithms in sex classification.
Population
Differences in skull structure among various populations greatly impact the accuracy of both metric and non-metric assessments for sex estimation. For instance, the study conducted by Techataweewan et al. [43] highlighted sexual dimorphic differences in skulls among Thai, American, and African populations. They measured 20 cranial metrics on 319 dried Thai skulls and compared the results to those reported by L’Abbé et al. for Americans and Africans. Across all craniometric characteristics, the average percentage differences between males and females were found to be 5.62% in Americans and 3.65% in Africans. The Thai population exhibited an intermediate difference of 4.78%. Furthermore, several non-metric studies indicated that Walker’s sex estimation standard inadequately represents the sexual dimorphism presented by several populations, resulting in incorrect sex estimation for many individuals [76, 87, 112–114]. Consequently, expanding the investigated populations and implementing population-specific adjustments is imperative to account for the varying patterns of sexual dimorphism.
Modalities
With a spurt of progress in medical imaging technology, the scope of materials available for sex estimation research has expanded. Medical imaging enables the broadening of databases from skull collections to larger hospital records, thereby encompassing characteristics from contemporary, dynamic populations. However, it is noteworthy that the use of hospital records simultaneously introduces ethical concerns surrounding the acquisition of medical records and data.
According to the literature we examined, CT is the most widely used medical imaging modality for skull-based sex estimation. Compared to X-rays, both CT and MRI provide great resolution without overlapping structural interference, but only one study used MRI technology [85]. In contrast to MRI, CT imaging has higher resolution and clarity for skeletal structures, allowing for clear observation of skull morphology, density, and subtle structural changes. This is particularly crucial for accurately measuring and analyzing skull features during sex estimation. Additionally, CT has higher scanning rates and is more cost-effective, making it better suited for forensic investigation and identification. More importantly, CT reconstruction technology has greatly facilitated precise 3D skull modeling.
The advent of CT reconstruction using the volume rendering technique provides a significant opportunity to model the skull collections at the 3D level. The measurements of 3D models and skull collections are relatively comparable due to the high resolution of CT scans producing negligible differences [63, 93, 121]. Moreover, digital 3D models provide flexible observation views around the entire skull and facilitate point positioning for operators. Combined with the aid of computer measuring software, 3D models enable researchers to explore more traits on the cranial surface. Consequently, sex estimation can benefit from 3D models, which allow detailed observation, precise landmarking, and computer-aided measurement, thereby enhancing the applicability and accuracy of both qualitative and quantitative methods [36–38, 63, 93, 122].
Non-metric and metric methods
The accuracy between non-metric and metric methods didn’t seem to exhibit a notable difference. However, one particular study by Jeong et al. [107] concluded that the metric methods for sex estimation are more useful than non-metric methods. They applied both traditional metric and non-metric methods for sex estimation on the mastoid process, revealing that the metrics achieved an accuracy rate 20% higher than the non-metrics. Consequently, the further comparison between metrics and non-metrics necessitates the standardization of variables, such as the same sample size, anatomical site, and predictive algorithm.
While traditional metric traits primarily reflect cranial size, GM methods offer a distinct perspective by capturing shape variations between male and female skulls. A study by Ajanovic et al. [89] employed GPA to investigate sex differences in the shape of orbital regions among Bosnians, exhibiting an average accuracy rate of 87.6%. This result is comparable to the 92.1% accuracy reported by Packirisamy et al. [123], who employed traditional metric indicators based on Saudi Arabian skulls.
In another study, Imaizumi et al. [116] created homologous surface models of the entire skull, cranium, and mandible using 100 skull shapes derived from CT reconstructions. They applied partial least squares regression to reduce dimensionality and generated principal components (PCs) for each model. When the first PC was input into an SVM classifier, the following accuracy rates were achieved: 90.6% for the whole skull, 90.7% for the cranium, and 84.1% for the mandible. The surface-based outcome mirrored the findings of several previous studies that employed traditional metric methods, indicating the efficacy of GM methods [36, 69, 94–96].
In terms of traditional metric traits, the bizygomatic breadth (BZB) has been found in studies from Turkey [73], Thailand [43], Iran [83], India [63, 68], Greece [119], and Caucasian populations [37] to be one of the most reliable predictors of sex, with an accuracy rate ranging from 77.2% to 89.2%. Additionally, as presented in relevant studies, the metric traits from partial regions of the skull could potentially be utilized for sex estimation.
Mandible.
Denny et al. [59] employed mandibular width for sex estimation among Indians, achieving an accuracy rate of 68.4%. However, Sain et al. [69] achieved a higher accuracy of 80% by utilizing the gonion-gnathion length derived from Indian mandibles. The ramus and chin height exhibited significant sexual dimorphism in both Rad et al.’s study [81] for Iranians and Cappella et al.’s study [124] for Italians. In addition to linear indicators, angular measurements of the mandible have also been studied for sex estimation. The condylar angle and notch angle were combined by Abualhija et al. [125], successfully classifying 77.6% of Jordanian samples. Meanwhile, Girdhar et al. [60] solely focused on the gonial angle and achieved a lower accuracy rate of 60.7% for sex estimation in the Indian population.
Mastoid.
The distance between the most prominent convex mastoid point and intermastoid distance were validated in the Iranian population, achieving an impressive accuracy rate of over 90% in two studies [79, 82]. In contrast, when measuring mastoid volumes in Korean individuals using 3D CT models, Jeong et al. [107] found a lower accuracy rate of 73.0%, similar to Petaros et al.’s 74.1% accuracy in Croatians [109]. Additionally, two Indian studies revealed 76.7% and 78.0% accuracy for mastoid height and area, respectively [66, 67].
Foramen magnum and occipital condyles.
Meral et al. [35] and Aljarrah et al. [126] reported that FM area was the most accurate single predictor, achieving an accuracy of 74.5% and 64.1%, respectively. In three separate studies [99, 102, 127], the FM width emerged as a significant sex difference, consistently achieving an accuracy of approximately 65.4%. In a Malaysian study, Soon et al. [128] obtained 60% accuracy using FM’s transverse and antero-posterior diameters. Additionally, the maximum length of the right OC was found to be significant for sex estimation in three investigations [99, 102, 127], with accuracy ranging from 58.9% to 66.5%. Notably, when combining the FM and OC predictors, a slight enhancement in the performance of sex estimation was observed in three studies [102, 126, 127].
Paranasal sinus.
Four studies employed 2-D CT scans and consistently reported that the maxillary sinus (MS) height was the most effective variable in indicating sexual dimorphism [58, 62, 65, 100]. Additionally, Kurniawan et al. [64] calculated the MS index from Indian lateral cephalograms, attaining an accuracy of 76.2%.
Banihashem et al. [80] analyzed the sphenoid sinus volume (SSV) using 2-D CBCT images from 469 Iranian individuals, demonstrating that the SSV could accurately distinguish sex with an average accuracy of 90.2%. In contrast, Liu et al. [85] achieved a lower accuracy of 63.3% when utilizing the SS area calculated from the midsagittal view of 73 Chinese MRI scans.
Two studies evaluated the effectiveness of the frontal sinus (FS) in sex estimation, achieving an approximate accuracy of 68.3% [50, 77]. They reported FS height, width, and anteroposterior diameters as useful indicators of sexual dimorphism.
Other anatomical regions.
Two studies indicated that the orbital traits had limited forensic value for sex classification in Croatian and Turkish populations, achieving an accuracy rate of approximately 73.0% [74, 108]. Conversely, Packirisamy et al. [123] demonstrated that the orbital region exhibited high potential for sex estimation among Saudi Arabians, achieving a high 92.1% accuracy rate. The orbit height was the most reliable variable, with an accuracy rate of 83.8%.
In terms of the crista galli (CG), Komut et al. [72] found that the CG length on axial slices of CT was the most effective variable, achieving an accuracy of 83.7%. Golpinar et al. [78], on the other hand, concluded that CG height was the most predictive factor, achieving an accuracy of 88.4%. Both studies underscored the significance of CG measurements in obtaining reliable sex classification results.
Furthermore, a study conducted by Uabundit et al. [118] on 124 dry skulls from Thailand suggested that pterion measurements might hold potential for sex estimation, achieving an accuracy of 80.7% in classification.
Analytical models
In the context of sex estimation, complex classical ML models do not necessarily achieve better accuracy. Sex estimation is a straightforward binary classification task. For this reason, simple statistical learning models are often capable of yielding satisfactory results. Consequently, when complex AI models were employed for sex estimation, they did not significantly enhance the accuracy rates. Additionally, the comparison between statistical models and AI algorithms was hampered by a limited number of studies (only 10 in total). This constraint might be another reason why the difference in accuracy between the two types of analytical models was relatively small.
Among the reviewed articles, four studies reported their accuracy rates for sex estimation exceeding 95%. Toneva et al. [93] conducted a comparative analysis of three models—SVM, NN, and LR—and found that SVM emerged as the most effective classifier, achieving an accuracy of 96.1%. In a separate study, Farhadian et al. [79] developed 11 models based on measurements from Iranian mastoids. Notably, the RF model achieved the highest accuracy rates, with 97.5% for training data and 96.9% for testing data, while statistical models lagged behind with an accuracy of 92.5%. Utilizing GM methods, Bertsatos et al. [117] generated elliptic Fourier descriptors from the 2D projection of the nasion–bregma ectocranial segment, which achieved an accuracy of 96.7% in the DFA model. A specific study by Kondou et al. [106] implemented a CNN algorithm using voxel data from 3D cranial models, achieving impressive accuracies of 95.0% and 93.0% for training and testing datasets, respectively.
These four studies, each showcasing a unique algorithm achieving high accuracy in sex estimation, underscore the variability in performance among different models. Therefore, the multi-angle evaluation of various algorithms is crucial for selecting the most efficient classifier [50]. By evaluating algorithms from multiple angles, researchers might gain a comprehensive understanding of their strengths and limitations, ultimately leading to more accurate and efficient predictive models.
Additionally, AI algorithms are often regarded as "black boxes," where their internal workings remain opaque to external scrutiny, raising ethical concerns regarding applying such algorithmic evidence in criminal justice proceedings. To address these concerns, future research is suggested to persist in exploring AI statistics for sex estimation, thus enhancing the interpretability of these algorithms [129, 130].
Software for skull-based sex estimation
The development and open access of software have improved the convenience of skull-based sex estimation, making the process more efficient and straightforward for users. A range of sex estimation software has become available. This includes non-metric tools like CroWalker [114] and MorphoPASSE [131], alongside traditional metric software such as SexEst [43], Fordisc [115], and KKU Sex Estimation [132], as well as GM applications like SkullProfiler [46] and 3D-ID [133]. This variety of software ensures that sex estimation from skulls is more comprehensive, offering practitioners a choice of tools suited to their specific needs. Furthermore, given the uniqueness of different populations, future software updates are expected to incorporate data from diverse worldwide populations to enhance the applicability and accuracy of these tools.
Limitations of the scoping review
Due to the multiple variables involved in sex estimation studies, such as population, sample size, modality, cranial traits, and prediction models, it becomes extremely challenging to precisely control experimental variables to assess the impact of one single factor. Therefore, our review only offers a rough outline of the efficacy differences among different technologies in sex estimation. Moreover, while DNA testing can achieve 100% accuracy in sex classification, our study doesn’t delve into genetic techniques. Conversely, our focus is primarily on the metric and non-metric methods widely employed in archaeology and anthropology. These methods not only assist DNA technique in initial sex estimation, but they also possess the unique ability to analyze cranial feature differences among various populations, thus offering valuable insights into human biological diversity.
Conclusions
Sex estimation has an extensive practical value in forensic anthropology. Our study employed modalities, methods, and models as a framework to investigate the strengths and limits of various technologies in skull-based sex estimation. Our purpose is to facilitate practitioners’ rapid comprehension of the components of sex estimation and empower them to select cranial features and research methodologies that optimally address their practical needs in the identification process. Additionally, in instances where extensive scientific research may be constrained, we summarized several software tools that have the potential to aid forensic investigators in efficiently and accurately performing sex estimation on skulls.
Through a synthesis of recent literature, we have discovered that with the development of medical 3D imaging, the efficiency of skull-based sex classification has been improved. However, despite these advancements, several limitations persist in this field, necessitating future attention. For instance, improving the generalization ability of models to handle skull data from diverse populations, reducing the technological costs to enable wider application scenarios, and validating established sex prediction models on independent datasets to ensure their robustness and applicability.
Looking ahead, research in sex estimation is expected to achieve breakthroughs in several areas. With the application of big data and AI techniques, it is anticipated that the performance of multiple models will be evaluated using different performance indicators, thereby enhancing the reliability of sex estimation outcomes. Furthermore, the combination of 3D imaging and computer vision techniques might capture more detailed information about skulls, providing finer data support for sex estimation.
Supporting information
S2 Table. Characteristics of the included articles.
https://doi.org/10.1371/journal.pone.0311762.s002
(DOCX)
References
- 1.
Christensen AM, Passalacqua NV, Bartelink EJ. Chapter 1—Introduction. In: Christensen AM, Passalacqua NV, Bartelink EJ, editors. Forensic Anthropology. San Diego: Academic Press; 2014. p. 1–17.
- 2. Kramer NA, Lopez-Capp TT, Michel-Crosato E, Biazevic MGH. Sex estimation from the mastoid process using Micro-CT among Brazilians: discriminant analysis and ROC curve analysis. J Forensic Radiol Im. 2018;14:1–7.
- 3. Krishan K, Chatterjee PM, Kanchan T, Kaur S, Baryah N, Singh RK. A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. Forensic Sci Int. 2016;261:165.e1–165.e8. pmid:26926105
- 4. Milella M, Franklin D, Belcastro MG, Cardini A. Sexual differences in human cranial morphology: Is one sex more variable or one region more dimorphic? Anat Rec. 2021;304(12):2789–810. pmid:33773067
- 5. Laffranchi Z, Charisi D, Jiménez-Brobeil SA, Milella M. Gendered division of labor in a Celtic community? A comparison of sex differences in entheseal changes and long bone shape and robusticity in the pre-Roman population of Verona (Italy, third-first century BC). Am J Phys Anthropol. 2020;173(3):568–88. pmid:32812226
- 6. Mello-Gentil T, Souza-Mello V. Contributions of anatomy to forensic sex estimation: focus on head and neck bones. Foren Sci Res. 2022;7(1):11–23. pmid:35341126
- 7. Cao YJ, Ma YG, Yang XT, Xiong J, Wang YH, Zhang JH, et al. Use of deep learning in forensic sex estimation of virtual pelvic models from the Han population. Foren Sci Res. 2022;7(3):540–9. pmid:36353321
- 8. Ives G, Johns SE, Deter C. Sexual dimorphism of pelvic scarring: A new method of adult biological sex estimation. J Forensic Sci. 2024;00:1–13. pmid:39036898
- 9. Zhang K, Zhan M, Deng L, Qiu LR Deng ZH. Estimation of stature and sex from pelvic measurements in a Chinese population. Aust J Forensic Sci. 2020;52(4):406–16.
- 10. Del Bove A, Menendez L, Manzi G, Moggi-Cecchi J, Lorenzo C, Profico A. Mapping sexual dimorphism signal in the human cranium. Sci Rep-UK. 2023;13(1):16847. pmid:37803023
- 11. Partido NM, Fombuena ZI, Borja MEA, Alemán AI. Discriminant functions for sex estimation using the rib necks in a Spanish population. Int J Legal Med. 2021;135(3):1055–65. pmid:33675384
- 12. Kubicka AM, Piontek J. Sex estimation from measurements of the first rib in a contemporary Polish population. Int J Legal Med. 2016;130(1):265–72. pmid:26289414
- 13. Peleg S, Kallevag RP, Dar G, Steinberg N, Masharawi Y, May H. New methods for sex estimation using sternum and rib morphology. Int J Legal Med. 2020;134(4):1519–30. WOS:000516298200001. pmid:32072241
- 14. Koşar Mİ, Uğuz Gençer C, Tetiker H, Yeniçeri İÖ, Çullu N. Sex and stature estimation based on multidetector computed tomography imaging measurements of the sternum in Turkish population. Foren Imag. 2022;28:200495.
- 15. Sehrawat JS. Sex estimation from discriminant function analysis of clavicular and sternal measurements: a forensic anthropological study based on examination of two bones of Northwest Indian subjects. Aust J Forensic Sci. 2018;50(1):20–41.
- 16. Garoufi N, Bertsatos A, Chovalopoulou ME, Villa C. Forensic sex estimation using the vertebrae: an evaluation on two European populations. Int J Legal Med. 2020;134(6):2307–18. pmid:32940842
- 17. Karaca AM, Senol E, Eraslan C. Evaluation of the usage of the cervical 7th vertebra in sex estimation with measurements on computerized tomography images. Legal Med-Tokyo. 2023;62:102220. pmid:36842227
- 18. Azofra-Monge A, Alemán Aguilera I. Morphometric research and sex estimation of lumbar vertebrae in a contemporary Spanish population. Forensic Sci Med Pat. 2020;16(2):216–25. pmid:32219709
- 19. Demir U, Etli Y, Hekimoglu Y, Kartal E, Keskin S, Yavuz A, et al. Sex estimation from the clavicle using 3D reconstruction, discriminant analyses, and neural networks in an Eastern Turkish population. Legal Med-Tokyo. 2022;56:102043. pmid:35183842
- 20. Bozdag M, Er A, Kranioti E, Basa CD, Oztop B, Kacmaz E, et al. Sex estimation in a modern Turkish population using the clavicle: a computed tomography study. Aust J Forensic Sci. 2022;54(2):187–98.
- 21. Hisham S, Lai PS, Ibrahim MA, Zainun KA. Sex estimation using post-mortem computed tomographic images of the clavicle in a Malaysian population. Legal Med-Tokyo. 2024;71:102500. pmid:39067245
- 22. Li Y, Niu CQ, Wang J, Xu Y, Dai H, Xiong T, et al. A fully automated sex estimation for proximal femur X-ray images through deep learning detection and classification. Legal Med-Tokyo. 2022;57:102056. pmid:35430525
- 23. Bertsatos A, Garoufi N, Chovalopoulou M-E. Advancements in sex estimation using the diaphyseal cross-sectional geometric properties of the lower and upper limbs. Int J Legal Med. 2021;135(3):1035–46. pmid:33029676
- 24. Phuwadon D. Sex estimation from upper limb bones in a Thai population. Anat Cell Biol. 2020;53(1):36–43. pmid:32274247
- 25. d’Oliveira Coelho J, Curate F. CADOES: An interactive machine-learning approach for sex estimation with the pelvis. Forensic Sci Int. 2019;302:109873. pmid:31382223
- 26. Stan E, Muresan C-O, Dumache R, Ciocan V, Ungureanu S, Costachescu D, et al. Sex Estimation from Computed Tomography of Os Coxae—Validation of the Diagnose Sexuelle Probabiliste (DSP) software in the Romanian population. Appl Sci-Basel. 2024;14(10):4136.
- 27. Qaq R, Mânica S, Revie G. Sex estimation using lateral cephalograms: A statistical analysis. Forensic Sci Int: Rep. 2019;1:100034.
- 28. Kimmerle EH, Ross A, Slice D. Sexual dimorphism in America: Geometric morphometric analysis of the craniofacial region. J Forensic Sci. 2008;53(1):54–7. pmid:18279240
- 29. Best KC, Garvin HM, Cabo LL. An investigation into the relationship between human cranial and pelvic sexual dimorphism. J Forensic Sci. 2018;63(4):990–1000. pmid:29044526
- 30. Profico A, Piras P, Buzi C, Di Vincenzo F, Lattarini F, Melchionna M, et al. The evolution of cranial base and face in Cercopithecoidea and Hominoidea: Modularity and morphological integration. Am J Primatol. 2017;79:e22721. pmid:29095513
- 31. Wellens HLL, Kuijpers-Jagtman AM, Halazonetis DJ. Geometric morphometric analysis of craniofacial variation, ontogeny and modularity in a cross-sectional sample of modern humans. J Anat. 2013;222(4):397–409. pmid:23425043
- 32. Patil KR, Mody RN. Determination of sex by discriminant function analysis and stature by regression analysis: a lateral cephalometric study. Forensic Sci Int. 2005;147(2–3):175–80. pmid:15567623
- 33. Ningtyas AH, Widyaningrum R, Shantiningsih RR, Yanuaryska RD. Sex estimation using angular measurements of nasion, sella, and glabella on lateral cephalogram among Indonesian adults in Yogyakarta. Egypt J Forensic Sci. 2023;13(1):48.
- 34. Luo HF, Wang JR, Zhang S, Mi CB. The application of frontal sinus index and frontal sinus area in sex estimation based on lateral cephalograms among Han nationality adults in Xinjiang. J Forensic Leg Med. 2018;56:1–4. pmid:29525579
- 35. Meral O, Belkıs Toklu B, Meydan R, Kaya A, Karadayı B, Acar T. Sex estimation from foramen magnum parameters in adult Turkish population: A computed tomography study. Legal Med-Tokyo. 2020;47:101775. pmid:32791469
- 36. Toneva DH, Nikolova SY, Agre GP, Zlatareva DK, Hadjidekov VG, Lazarov NE. Data mining for sex estimation based on cranial measurements. Forensic Sci Int. 2020;315:11044. pmid:32781389
- 37. Cekdemir YE, Mutlu U, Karaman G, Balci A. Estimation of sex using morphometric measurements performed on cranial computerized tomography scans. Radiol Med. 2021;126(2):306–15. pmid:32533549
- 38. Gao HJ, Geng GH, Yang W. Sex Determination of 3D Skull Based on a Novel Unsupervised Learning Method. Comput Math Method M. 2018;2018(1):4567267. pmid:30046351
- 39. Walker PL. Sexing skulls using discriminant function analysis of visually assessed traits. Am J Phys Anthropol. 2008;136(1):39–50. pmid:18324631
- 40. Haas J, Buikstra JE, Ubelaker DH, Aftandilian D, History FMoN, Survey AA. Standards for Data Collection from Human Skeletal Remains. America: Arkansas Archeological Survey; 1994.
- 41. Bu WQ, Guo YX, Zhang D, Du SY, Han MQ, Wu ZX, et al. Automatic sex estimation using deep convolutional neural network based on orthopantomogram images. Forensic Sci Int. 2023;348:111704. pmid:37094502
- 42. Braun S, Ridel AF, L’Abbé EN, Theye CEG, Oettlé AC. Repeatability of a morphoscopic sex estimation technique for the mental eminence on micro-focus X-ray computed tomography models. Foren Imag. 2022;28:200500.
- 43. Techataweewan N, Hefner JT, Freas L, Surachotmongkhon N, Benchawattananon R, Tayles N. Metric sexual dimorphism of the skull in Thailand. Forensic Sci Int: Rep. 2021;4:100236.
- 44. Madadin M, Menezes RG, Al Saif HS, Abu Alola H, Al Muhanna A, Gullenpet AH, et al. Morphometric evaluation of the foramen magnum for sex determination: A study from Saudi Arabia. J Forensic Leg Med. 2017;46:66–71. pmid:28157592
- 45. Webster M, Sheets HD. A practical introduction to landmark-based geometric morphometrics. The Paleontological Society Papers. 2010;16:163–88.
- 46. Caple JM, Byrd JE, Stephan CN. The utility of elliptical Fourier analysis for estimating ancestry and sex from lateral skull photographs. Forensic Sci Int. 2018;289:352–62. pmid:29957511
- 47. Cechová M, Dupej J, Bruzek J, Bejdova S, Horák M, Velemínská J. Sex estimation using external morphology of the frontal bone and frontal sinuses in a contemporary Czech population. Int J Legal Med. 2019;133(4):1285–94. pmid:30982130
- 48.
Yang W, Liu XN, Zhu F, Geng GH, Li K. Determination of sex discriminant function analysis in Chinese human skulls. In Zhou J, Wang YH, Sun ZN, Jia ZH, Feng JJ, Shan SG, et al. editors. Biometric Recognition. 13th Chinese Conference on Biometric Recognition (CCBR); 2018 Aug 11–12; Urumqi, China. Springer Cham; 2018. p. 589–98. https://doi.org/10.1007/978-3-319-97909-0_63
- 49. Toneva DH, Nikolova SY, Zlatareva DK, Hadjidekov VG, Lazarov NE. Sex estimation by mastoid triangle using 3D models. Homo. 2019;70(1):63–73. pmid:31475292
- 50. Navega D, Vicente R, Vieira DN, Ross AH, Cunha E. Sex estimation from the tarsal bones in a Portuguese sample: a machine learning approach. Int J Legal Med. 2015;129(3):651–9. pmid:25186617
- 51. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521(7553):436–44. Epub 2015/05/29. pmid:26017442
- 52. Yamashita R, Nishio M, Do RKG, Togashi K. Convolutional neural networks: an overview and application in radiology. Insights Imaging. 2018;9(4):611–29. pmid:29934920
- 53. Breit HC, Varga-Szemes A, Schoepf UJ, Emrich T, Aldinger J, Kressig RW, et al. CNN-based evaluation of bone density improves diagnostic performance to detect osteopenia and osteoporosis in patients with non-contrast chest CT examinations. Eur J Radiol. 2023;161:110728. pmid:36773426
- 54. Fu BK, Zhang MD, He JJ, Cao Y, Guo YC, Wang RP. StoHisNet: A hybrid multi-classification model with CNN and transformer for gastric pathology images. Comput Meth Prog Bio. 2022;221:106924. pmid:35671603
- 55. Shah A, Shah M, Pandya A, Sushra R, Sushra R, Mehta M, et al. A comprehensive study on skin cancer detection using artificial neural network (ANN) and convolutional neural network (CNN). Clinical eHealth. 2023;6:76–84.
- 56. Gour N, Khanna P. Multi-class multi-label ophthalmological disease detection using transfer learning based convolutional neural network. Biomed Signal Proces. 2021;66:102329.
- 57. Tricco AC, Lillie E, Zarin W, O’Brien KK, Colquhoun H, Levac D, et al. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Ann Intern Med. 2018;169(7):467–73. pmid:30178033
- 58. Deshpande AA, Munde AD, Mishra SS, Kawsankar KD, Sawade RV, Mandar B. Determination of sexual dimorphism of maxillary sinus using cone-beam computed tomography in a rural population of western Maharashtra- A retrospective, cross-sectional study. J Family Med Prim Care. 2022;11(4):1257–61. pmid:35516696
- 59. Denny C, Bhoraskar M, Abdul Aziz Shaikh S, T S B, Sujir N, Natarajan S. Investigating the link between frontal sinus morphology and craniofacial characteristics with sex: A 3D CBCT study on the South Indian population. F1000Research. 2023;12:811. pmid:37744763
- 60. Girdhar A, Keerthika R, Narwal A, Kamboj M, Devi A, Sharma R. Comparative manual and digital analysis of gonial angle in lateral cephalograms for gender determination. Forensic Sci Med Pat. 2024;20(1):73–8. pmid:37060537
- 61. Jacob M, Bindhu S. Morphometric assessment methods for estimation of sexual dimorphism from cranial occipital condyles. Int J Morphol. 2022;40(4):1128–33.
- 62. Kannampurath A, Srikantannair SL, Mathew P, Sivaprasad T. Maxillary sinus in gender determination: a morphometric analysis using cone beam computed tomography. Forensic Sci Med Pathol. 2023 Nov 12. pmid:37952242
- 63. Kumar Battan S, Sharma M, Gakhar G, Garg M, Singh P, Jasuja OP. Cranio-facial bones evaluation based on clinical CT data for sex determination in Northwest Indian population. Legal Med-Tokyo. 2023;64:102292. pmid:37392574
- 64. Kurniawan A, Athalia SA, Prakoeswa BFWR, Rizky BN, Chusida A, Yudha M, et al. Cephalometric radiograph-based approach for sex determination using maxillary sinus index in Surabaya, Indonesia. Eur J Anat. 2023;27(6):663–8.
- 65. Mathew A, Jacob L. 3D evaluation of maxillary sinus in gender determination: A cone beam computed tomography study. J Indian Acad Oral M. 2020;32(4):384–9.
- 66. Mondal B, Vaishali MR, David MP, Roopashri G, Kumar V, Ponnuswamy IA. Assessment of the usefulness of morphometric and volumetric analysis of mastoid process for gender determination in forensic odontology: A retrospective cone beam computed tomographic study. J Indian Acad Oral M. 2022;34(1):82–6.
- 67. Passey J, Pandey S, Passey N, Singh R, Singh R, Kumar A. Radiographic Evaluation of Mastoid Parameters for Sexual Differentiation in North Indian Population. Cureus. 2021;13(6):e16011. pmid:34336501.
- 68. Ramamoorthy B, Pai MM, Ullal S, Prabhu LV. Discriminant function analysis of craniometric traits for sexual dimorphism and its implication in forensic anthropology. J Anat Soc India. 2019;68(4):260–8.
- 69. Saini V, Chowdhry A, Mehta M. Sexual dimorphism and population variation in mandibular variables: a study on a contemporary Indian population. Anthropol Sci. 2022;130(1):59–70.
- 70. Sathawane R, Tripathi S, Sukhadeve V. Introduction of a new parameter ’Mastoid to Gonion Height’ and its comparative evaluation with established pa cephalometric parameters in sex determination: An analytical study. J Indian Acad Oral M. 2022;34(2):218–22.
- 71. Verma P, Gupta N, Sameera Y, Faraz S, Sharma P, Sharma B. Foramen magnum as determinant of sexual dimorphism in Sri Ganganagar population: A radiographic study. J Indian Acad Oral M. 2021;33(1):71–6.
- 72. Komut E, Golpinar M. A comprehensive morphometric analysis of crista galli for sex determination with a novel morphological classification on computed tomography images. Surg Radiol Anat. 2021;43(12):1989–98. pmid:34245351
- 73. Meral O, Meydan R, Toklu BB, Kaya A, Karadayi B, Acar T. Estimation of sex from computed tomography images of skull measurements in an adult Turkish population. Acta Radiol. 2022;63(11):1513–21. pmid:34623180
- 74. Meral O, Toklu BB, Meydan R, Kaya A, Karadayı B, Acar T. Sexing from the orbital measurements using computed tomography images and discriminant function analysis in Turkish population. Imaging Sci J. 2022;70(4):207–13.
- 75. Toy S, Secgin Y, Oner Z, Turan MK, Oner S, Senol D. A study on sex estimation by using machine learning algorithms with parameters obtained from computerized tomography images of the cranium. Sci Rep-UK. 2022;12(1):4278. pmid:35277536
- 76. Yaşar B, Sağır M. Testing of morphological sex estimation traits with a sex-known collection: Ottoman period skulls. Int J Osteoarchaeol. 2023;33(6):1042–51.
- 77. Emekli E. Sex determination using frontal sinus diameters on direct radiography. Cureus. 2023;15(10):e47476. pmid:38022337.
- 78. Golpinar M, Salim H, Ozturk S, Komut E, Sindel M. Sex estimation with morphometric and morphological characteristics of the crista galli. Surg Radiol Anat. 2022;44(7):1007–15. pmid:35750936
- 79. Farhadian M, Salemi F, Shokri A, Safi Y, Rahimpanah S. Comparison of data mining algorithms for sex determination based on mastoid process measurements using cone-beam computed tomography. Imagng Sci Dent. 2020;50(4):323–30. pmid:33409141
- 80. Banihashem Rad SA, Anbiaee N, Moeini S, Bagherpour A. Sex determination using human sphenoid sinus in a northeast Iranian population: A discriminant function analysis. J Dent. 2023;24:95–102. pmid:37051499
- 81. Rad FO, Javanshir B, Nemati S, Khaksari F, Mansoori R, Ranjzad H, et al. Evaluation of sexual dimorphism with mandibular parameters by digital panoramic radiography. Open Dent J. 2020;14(1):172–7.
- 82. Salemi F, Farhadian M, Shokri A, Safi Y, Rahimpanah S. Sex determination by osteometric assessment of the mastoid process using Cone Beam Computed Tomography. Brazilian Dental Science. 2021;24(1):1–9.
- 83. Salmanipour A, Memarian A, Tofighi S, Vahedifard F, Khalaj K, Shiri A, et al. Prediction of sex, based on skull CT scan measurements in Iranian ethnicity by machine learning-based model. Foren Imag. 2023;33:200549.
- 84. Sobhani F, Salemi F, Miresmaeili A, Farhadian M. Morphometric analysis of the inter-mastoid triangle for sex determination: Application of statistical shape analysis. Imagng Sci Dent. 2021;51:1–8. pmid:34235062
- 85. Liu HY, Bai ZY, Jing W, Yang XF, Kwon TG, Yun KM, et al. Application of sphenoid sinus morphological characteristics in sex estimation based on magnetic resonance imaging images. Int J Morphol. 2023;41(4):1166–70.
- 86. Yang W, Zhou MQ, Zhang PF, Geng GH, Liu XN, Zhang HB. Skull sex estimation based on wavelet transform and Fourier transform. Biomed Res Int. 2020;2020(1):8608209. pmid:32420376
- 87. Yang XT, Sun CH, Ma YG, Cao YJ, Xiong J, Zhang J, et al. Sex estimation of Han adults in Western China based on three-dimensional cranial CT reconstruction. Journal of Forensic Medicine. 2023;39(1):27–33. pmid:37038852
- 88. Aida SH, Zurifa A, Ilvana H, Senad S, Maida RT, Izeta A, et al. Bioanthropological analysis of human occipital condyles using geometric morphometric method. Saudi J Biol Sci. 2020;27(12):3415–20. pmid:33304150
- 89. Ajanović Z, Ajanović U, Dervišević L, Hot H, Voljevica A, Talović E, et al. A geometric morphometrics approach for sex estimation based on the orbital region of human skulls from Bosnian population. Scanning. 2023;2023(1):2223138. pmid:37089258
- 90. Ajanovic Z, Dervisevic L, Dervisevic A, Sarac-Hadzihalilovic A, Dervisevic E, Tokic JB, et al. Sex prediction by geometric morphometric analysis of the hard palate. Eur Rev Med Pharmaco. 2022;26(17):6057–64. pmid:36111905
- 91. Ajanovic Z, Dervisevic L, Dervisevic E, Lujinovic A, Ajanovic U, Biscevic-Tokic J, et al. Sex estimation based on foramen magnum: A three-dimensional geometric morphometrics approach. Int J Morphol. 2023;41(2):410–6.
- 92. Sarač-Hadžihalilović A, Ajanović Z, Hasanbegović I, Šljuka S, Rakanović-Todić M, Aganović I, et al. Analysis of gender differences on pyriform aperture of human skulls using geometric morphometric method. Folia Morphol. 2022;81(3):707–14. pmid:34608986.
- 93. Toneva D, Nikolova S, Agre G, Zlatareva D, Hadjidekov V, Lazarov N. Machine learning approaches for sex estimation using cranial measurements. Int J Legal Med. 2021;135(3):951–66. pmid:33179173
- 94. Toneva D, Nikolova S, Tasheva-Terzieva E, Zlatareva D, Lazarov N. A geometric morphometric study on sexual dimorphism in viscerocranium. Biology-Basel. 2022;11(9):1333. pmid:36138812
- 95. Toneva DH, Nikolova SY, Fileva NF, Zlatareva DK. Size and shape of human mandible: Sex differences and influence of age on sex estimation accuracy. Legal Med-Tokyo. 2023;65:102322. pmid:37722156
- 96. Toneva DH, Nikolova SY, Tasheva-Terzieva ED, Zlatareva DK, Lazarov NE. Sexual dimorphism in shape and size of the neurocranium. Int J Legal Med. 2022;136(6):1851–63. pmid:35945460
- 97. Boucherie A, Chapman T, García-Martínez D, Polet C, Vercauteren M. Exploring sexual dimorphism of human occipital and temporal bones through geometric morphometrics in an identified Western-European sample. Am J Biol Anthropol. 2022;178(1):54–68.
- 98. Jaafar A, Tersoo MA, Farrau U, Aliyu IS, Adamu LH, Ibrahim MZ, et al. The use of foramen magnum in the determination of sex using Computed Tomography (CT) scan images of sampled population attending National Ear Care Centre Kaduna State, Nigeria. Arab Journal of Forensic Sciences and Forensic Medicine. 2022;4(2):112–26.
- 99. Ominde BS, Igbigbi PS. A retrospective study to evaluate the morphometry of the foramen magnum and its role in forensic science in a nigerian population of Delta State. Journal of Forensic Science and Medicine. 2022;8(2):46–51.
- 100. Ominde BS, Ikubor JE, Iju WJ, Ebeye A, Igbigbi PS. Morphometry of the maxillary sinus and its role in sex determination in Delta State Nigeria: Retrospective CT study. Journal of Indian Academy of Forensic Medicine. 2023;45(1):52–7.
- 101. Meinerová T, Sutoová D, Havelková PB, Velemínská J, Dupej J, Bejdová S. How reliable is the application of the sex classifier based on exocranial surface (Musilova et al., 2016) for geographically and temporally distant skull series. Forensic Sci Int. 2023;352:111850. pmid:37827023
- 102. Abo El-Atta HMH, Abdel-Rahman RH, El-Hawary G, Abo El-Al-Atta HM. Sexual dimorphism of foramen magnum: An Egyptian study. Egypt J Forensic Sci. 2020;10(1):1.
- 103. Sassi C, Picapedra A, Álvarez-Vaz R, Schmidt CM, Ulbricht V, Júnior ED, et al. Sex determination in a brazilian sample from cranial morphometric parameters-a preliminary study. Journal of Forensic Odonto-Stomatology. 2020;38(1):8–17. pmid:32420908
- 104. Alves N, Deana NF, González J, Hernández P, Ceballos F. Sex prediction by analysis of the morphological characteristics of macerated skulls. Int J Morphol. 2020;38(4):815–9.
- 105. Berezowski V, Rogers T, Liscio E. Evaluating the morphological and metric sex of human crania using 3-dimensional (3D) technology. Int J Legal Med. 2021;135(3):1079–85. pmid:32382975
- 106. Kondou H, Morohashi R, Kimura S, Idota N, Matsunari R, Ichioka H, et al. Artificial intelligence-based forensic sex determination of East Asian cadavers from skull morphology. Sci Rep-UK. 2023;13(1):21026. pmid:38030742
- 107. Jeong YH, Koo HN, Kim YS, Lee B, Kim S, Shim YT. Using 3D images of Korean’s mastoid process to estimate sex: A metric study. Foren Imag. 2022;31:200527.
- 108. Kresic E, Basic Z, Jerkovic I, Kruzic I, Cavka M, Erjavec I. Sex estimation using orbital measurements in the Croatian population. Forensic Sci Med Pat. 2023;19(3):303–9. pmid:36151406
- 109. Petaros A, Sholts SB, Čavka M, Šlaus M, Wärmländer SKTS. Sexual dimorphism in mastoid process volumes measured from 3D models of dry crania from mediaeval Croatia. HOMO. 2021;72(2):113–27. pmid:33846705
- 110. Čechová M, Dupej J, Brůžek J, Bejdová Š, Velemínská J. A test of the Bulut et al. (2016) landmark-free method of quantifying sex differences in frontal bone roundness in a contemporary Czech sample. J Forensic Sci. 2021;66(2):694–9. pmid:33104239
- 111. Taek Shim Y, Kim DI, Aum N, Gyu Choi S, Lee YS, Koo HN, et al. Statistical classification methods for estimating sex based on five skull traits: A nonmetric assessment using 3D CT models. HOMO. 2023;74(1):45–54. pmid:36920118
- 112. Rogers LM, Halcrow SE, Kleffmann T, King CL. Comparing Walker’s (2008) skull trait sex estimation standard to proteomic sex estimation for a group of South Asian individuals. Forensic Science International: Synergy. 2024;8:100450. pmid:38314398
- 113. Cappella A, Bertoglio B, Di Maso M, Mazzarelli D, Affatato L, Stacchiotti A, et al. Sexual dimorphism of cranial morphological traits in an Italian sample: A population-specific logistic regression model for predicting sex. Biology-Basel. 2022;11(8):1202. pmid:36009828
- 114. Bareša T, Jerković I, Jerković N, Dolić K, Dujić G, Borić MĆ, et al. Walker’s traits for sex estimation in modern Croatian population using MSCT virtual cranial database: Validation and development of population-specific standards. Foren Imag. 2024;36:200578.
- 115. Nikita E, Nikitas P. On the use of machine learning algorithms in forensic anthropology. Legal Med-Tokyo. 2020;47:101771. pmid:32795933
- 116. Imaizumi K, Bermejo E, Taniguchi K, Ogawa Y, Nagata T, Kaga K, et al. Development of a sex estimation method for skulls using machine learning on three-dimensional shapes of skulls and skull parts. Foren Imag. 2020;22:200393.
- 117. Bertsatos A, Chovalopoulou ME, Brůžek J, Bejdová Š. Advanced procedures for skull sex estimation using sexually dimorphic morphometric features. Int J Legal Med. 2020;134(5):1927–37. pmid:32504147
- 118. Uabundit N, Chaiyamoon A, Iamsaard S, Yurasakpong L, Nantasenamat C, Suwannakhan A, et al. Classification and morphometric features of pterion in thai population with potential sex prediction. Medicina-Lithuania. 2021;57(11):1282. pmid:34833500
- 119. Nikita PA, Garoufi N, Valakos E, Constantinou C, Nikita E, Chovalopoulou ME. Testing the accuracy of the SexEst software for sex estimation in a modern Greek sample. Int J Osteoarchaeol. 2024;34(2):e3283.
- 120. Constantinou C, Nikita E. SexEst: An open access web application for metric skeletal sex estimation. Int J Osteoarchaeol. 2022;32(4):832–44.
- 121. Hamdan S, Rahmat RA, Razak FA, Kadir KAA, Abdullah E, Ibrahim N. Sex estimation of Malaysian sub-adults using craniometrics: A computed tomography study. Legal Med-Tokyo. 2023;64:102275. pmid:37229938
- 122. Yang W, Liu XN, Wang KG, Hu JB, Geng GH, Feng J. Sex determination of three-dimensional skull based on improved backpropagation neural network. Comput Math Method M. 2019;2019(1):9163547. pmid:30774706
- 123. Packirisamy V, Aljarrah K, Nayak SB. Morphometric evaluation of the orbital region for sex determination in a Saudi Arabian population using 3D CT images. Anat Sci Int. 2024;99(1):118–26. pmid:37721654
- 124. Cappella A, Gibelli D, Vitale A, Zago M, Dolci C, Sforza C, et al. Preliminary study on sexual dimorphism of metric traits of cranium and mandible in a modern Italian skeletal population and review of population literature. Legal Med-Tokyo. 2020;44:101695. pmid:32259691
- 125. Abualhija D, Revie G, Manica S. Mandibular ramus as a sex predictor in adult Jordanian subjects. Foren Imag. 2020;21:200366.
- 126. Aljarrah K, Packirisamy V, Al Anazi N, Nayak SB. Morphometric analysis of foramen magnum and occipital condyle using CT images for sex determination in a Saudi Arabian population. Morphologie. 2022;106(355):260–70. pmid:34391659
- 127. Atreya A, Shrestha R, Bhandari K, Malla SK, Acharya S, Menezes RG. Morphometric analysis of the foramen magnum in sex estimation: An additional 3DCT study from Nepal on a larger sample. Health Sci Rep-US. 2023;6(1):e999. pmid:36544619
- 128. Soon LP, Yi-Li W, Noor MHM, Ibrahim MA. Osteometry of foramen magnum by using post-mortem computed tomography (pmct) for discriminant analysis of sex and population affinity among Malaysian population. Arab Journal of Forensic Sciences and Forensic Medicine. 2023;5(1):43–57.
- 129. Ferrell MJ, Schultz JJ, Adams DM. Sex estimation research trends in forensic anthropology between 2000 and 2022 in five prominent journals. J Forensic Sci. 2024;69(4):1138–54. pmid:38600623
- 130. Tournois L, Trousset V, Hatsch D, Delabarde T, Ludes B, Lefèvre T. Artificial intelligence in the practice of forensic medicine: a scoping review. Int J Legal Med. 2024;138(3):1023–37. pmid:38087052
- 131.
Klales AR. Chapter 16—MorphoPASSE: Morphological pelvis and skull sex estimation program. In: Klales AR, editor. Sex Estimation of the Human Skeleton: Academic Press; 2020, p. 271–8.
- 132. Manthey L, Jantz RL, Vitale A, Cattaneo C. Population specific data improves Fordisc®’s performance in Italians. Forensic Sci Int. 2018;292:263.e1–.e7. pmid:30342922
- 133.
Slice D (Florida State University, Department of Scientific Computing, Tallahassee, Florida), Ross A (NC State University, Department of Biological Sciences, Raleigh, North Carolina). Geometric morphometric classification of crania for forensic scientists. Final report. National Institute of Justice, U.S. Department of Justice; 2009 Jul. Report No.: 2005-MU-BX-K078.