Morphometric and volumetric analysis of the frontal sinus in a Brazilian population using Cone beam Computed Tomography: a forensic approach for sex, age and facial morphology estimation

JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 ABSTRACT The aim of this study was to evaluate the linear and volumetric measurements of the frontal sinus (FS), using cone beam computed tomography (CBCT) scans, for a discriminatory analysis of gender, age and facial skeletal pattern in a Brazilian population. A total of 300 CBCT scans were analyzed, measuring the height, width, length and volume of the FS. The measurements were divided into groups: sex, age (20-24.9, 25-29.9, 30-34.9, 35-39.9, 40-45.9, 46-49.9, 50-54.9, 55-59.9, and 60-64.9 years) and facial skeletal pattern classes I, II and III. The function values in the centroid group were 0.675 for male and -0.292 for female. A rule was established indicating that if the value of D is greater than 0.19, the sample will be classified as male. The results showed a significant difference in women, who had significantly lower volume, width and depth than men ( p<0.001, p=0.003, and p<0.001, respectively). No significant differences could be observed between the age and facial skeletal pattern groups. The results suggest that the FS measures of volume, height, width and depth have moderate discriminatory power for predicting gender in a Brazilian population. In conclusion, the results show that the FS has potential for assessing gender, but the accuracy of the method and its applicability for analyzing age and facial skeletal pattern were limited in our population. INTRODUCTION Human identification plays a fundamental role humanitarianly and in procedural law, especially in civil and criminal cases. In the case of criminal investigations, natural disasters, plane crashes, cases of disappearances,²,³ where it is not possible to apply conventional identification methods, the analysis of images of the anatomical structures of the skull becomes extremely important.4 Because the skull, in addition to being considered the second best bone structure for age estimation, has the advantage of being highly resistant to damage.4 A particularly relevant anatomical structure in the cranial region is the frontal sinus (FS). It is characterized by a great variety and asymmetry and is unique to each individual.4 Because it is protected by skull bones, it is resistant to trauma, presenting a high probability of remaining intact during catastrophes and mass accidents, thus allowing the identification of individuals and sex estimation. 5 In addition to sex estimation, assessing age and skeletal class are relevant aspects. 21 Copyright © 2025 International Organization for Forensic Odonto-Stomatology - IOFOS Letícia Bego de Miranda1, Beatriz Caio Felipe2, Matheus Kawana Couto1, Wilton Mitsunari Takeshita3, Lilian Cristina Vessoni Iwaki¹, Mariliani Chicarelli da Silva¹ 1Department of Dentistry, State University of Maringá - Maringá, Parana, Brazil, ²Department of Oral Diagnosis, Piracicaba Dental School, State University of Campinas - Piracicaba, Sao Paulo, Brazil, ³Departament of Surgery and Diagnosis, Araçatuba Dental School, Sao Paulo State University, Araçatuba, Sao Paulo, Brazil. Corresponding author: [email protected] The authors declare that they have no conflict of interest. KEYWORDS Computed tomography, Frontal sinus, Forensic identification, Forensic dentistry. J Forensic Odontostomatol 2025. Aug;(43): 2 -21:31 ISSN":2219-6749 DOI: doi.org/10.5281/zenodo.15564344 Morphometric and volumetric analysis of the frontal sinus in a Brazilian population using Cone beam Computed Tomography: a forensic approach for sex, age and facial morphology estimation JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Globally, the literature is scarce in studies that explore the FS using cone beam computed tomography (CBCT) to estimate age,6 sex or facial skeletal pattern. Although this topic has been little studied in the literature so far, only a limited number of studies addressing sex were found7-9 and none of them investigated its relationship with forensic identification. Even though there are investigations in the area, to our knowledge, only two studies have evaluated the FS as an anatomical repair that can be used in human identification in a Brazilian population.10,11 The current literature regarding CBCT has reinforced the significant distinction in the morphological characteristics of FS in relation to sex.1,4,12,13 However, there are studies that question this reliability of the FS with identification methods,14-17 many of which employ bidimensional (2D) scans and in some cases, resort to the use of dry skulls. According to what has been studied in the literature up to date, there have been few volumetric analyses of the FS in relation to the age of adult individuals, which showed an expansion up to 40 years of age and a tendency to decrease with aging.18,19 Regarding the relationship between the FS and facial skeletal pattern, there are divergences in the literature. While Sawada et al. (2022) found no significant association, these results contrast with previous studies.7-9 Therefore, the objective of this work was to evaluate the linear and volumetric measurements of the FS using CBCT, in order to establish a discriminative functional analysis for estimating sex, age and facial skeletal pattern in a Brazilian population. This study aims to fill a significant gap in the literature by providing valuable stimuli for the practical application of FS in forensic science by contributing to forensic databases. MATERIALS AND METHODS This work was sent and approved by the Permanent Ethics Committee for Research Involving Human Beings at the State University of Maringá (UEM) under number CAAE: 68966523.2.0000.0104. Due to the retrospective nature of this study, signed informed consent was not required by the Committee. Sample selection A total of 300 CBCT scans from anonymous Brazilian individuals were included in the research, divided into the following groups: sex, age (20-24.9, 25-29.9, 30-34.9, 35-39.9, 40-45.9, 46-49.9, 50-54.9, 55-59.9, and 60-64.9) and facial skeletal pattern (Class I (0 ° < ANB < 4°), Class II (ANB ≥ 4°) and Class III (ANB ≤ 0°) using Steiner classification.19,20 The used scans belong to the archive of the Clinical Imaging Research Laboratory of the State University of Maringá (UEM) and were carried out as an indication for the most varied dental treatments. The exclusion criteria were malformations, obliterations, pathologies and diseases, trauma, absence of the frontal sinus on one side and patients under 20 years of age.21 As a result, 64 scans were excluded from the sample, therefore resulting in 235 scans to be analyzed. The final sample distribution is represented by tables 1 and 2. Table 1. Sample division according to sex and age Group Male Female Total 20-24.9 23 37 60 25-29.9 11 39 50 30-34.9 7 13 20 35-39.9 7 19 26 40-45.9 6 13 19 46-49.9 6 19 25 50-54.9 5 11 16 55-59.9 4 6 10 60-64.9 3 6 9 22 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Table 2. Sample division according to sex and facial skeletal pattern Image acquisition The scans were performed at the Clinical Imaging Research Laboratory (LIPC) of the Health Technology Center (CTS), of the Research Support Center Complex (COMCAP), located in the Department of Dentistry of the State University of Maringá (DOD/UEM) by the same oral and maxillofacial radiology team and are archived. They were obtained using the i-CAT Next Generation® equipment (Imaging Sciences International, Hatfield, PA, USA), with a volume of 300µ isometric voxel, FOV (Field of View) of 17 × 23 cm, tube voltage of 120 kVp and tube current of 3-8 mA. For the analysis, the Dolphin 3D Imaging software (Dolphin Imaging & Management Solutions®, Chatsworth, CA, USA) version 11.9 was used. Image analysis For FS analyses, CBCT scans were exported using the Digital Imaging extension in Communications in Medicine (DICOM) and were imported into the Dolphin 3D Imaging software (Dolphin Imaging & Management Solutions®, Chatsworth, CA, USA), version 11.9. For each scan, the reconstructions were oriented, positioning the Frankfurt plane (line that passes horizontally from the upper edge of the external auditory meatus to the lower orbital edge) parallel to the horizontal plane in the sagittal reconstruction before measurements. The evaluators underwent calibration through 10 CBCT scans. These assessments took place individually, randomly and in a dark and silent environment. Calibration lasted two weeks to ensure its reliability. The results were compared and discussed among the evaluators before the official measurements began. Linear and volumetric measurements were performed on 236 CBCT scans by two independent oral and maxillofacial radiologists (with more than 5 years of experience), blind to details of age, sex and facial skeletal class. To avoid eye fatigue, the two examiners evaluated only 10 scans per day.In order to assess sex, age and facial skeletal class, the following measurements were defined: height, width and length. These linear measurements (height, width and length) were taken in the coronal and axial reconstructions in the widest area of the FS13, and the volume of the FS was acquired by delimiting its anatomy in the three reconstructions (axial, coronal, and sagittal), both using the Dolphin® software.22,23 In the coronal reconstruction, the width was measured as the greatest distance between the medial and lateral walls of the FS, and the height as the greatest distance between the FS floor and ceiling (Figure 1). The depth was measured in the axial reconstruction as the distance from the anterior wall of the sinus to the posterior wall (Figure 2).13 To analyze the FS volume, the sinus/airway tool of the Dolphin® software was used as it allows reliable measurements, which produces a complete filling of the delimited region, preventing differences in area from interfering with the reliability of the data.23 In order to correctly segment the entire FS, regardless of its anatomical variability, the first step was to delimit the region of interest with points (green dots) that consequently formed a line, on its anatomical border in the coronal reconstruction (Figure 3). Afterwards, the exact delimitations in the other planes were checked (Figures 4 and 5), if any point was outside the delimitation, it could be changed.23 After this, “seed points” (yellow dots) were inserted in the frontal sinus, whose function is to expand within the airway, up to the selected limit, and when necessary, more “seed points” can be added. Sex Facial Skeletal Pattern Class I Class II Class III Male n=35 n=17 n=20 Female n=54 n=68 n=43 Total n=89 n=83 n=63 23 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Figure 1. Linear dimensions for supero-inferior (vertical line) and medio-lateral (horizontal line) length of the FS Figure 2. Linear dimension of the antero-posterior length of the FS Before the final volumetric calculation, the sensitivity threshold was also adjusted in all image acquisitions, being a tool that controls the filling of the volume of the region of interest.24 Although the sensitivity threshold is automatically determined by the Dolphin® software,25 for the present study the average value was ±50, adapted for the FS. After complete segmentation of the region of interest, the Dolphin® software automatically displayed the volume in cubic millimeters (mm³) of the selected area and a three-dimensional (3D) model of the FS (Figure 6). All viewing plans were checked to ensure that the enclosed area had been completely filled.27 All images were analyzed following the established sequence protocol: linear measurements, volumetric analysis and sex, age and facial skeletal class comparison with the patient’s known data. 24 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Figure 3. Delimitation of the anatomical borders of the FS in coronal reconstruction Figure 4. Delimitation of the anatomical borders of the FS in sagittal reconstruction Figure 5. Delimitation of the anatomical borders of the frontal sinus in axial reconstruction 25 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Figure 6. 3D model generated by the Dolphin 3D Imaging software Statistical analysis A database for qualitative and quantitative variables was organized to allow tabulation and statistical analysis. All statistical procedures were calculated using the statistical programs SPSS 25.0 (SPSS Inc., Chicago, IL, USA) and Bioestat 5.3 (Instituto Mamirauá, Pará, Brazil). Initially, a descriptive statistical analysis was carried out to obtain absolute and relative numbers. The Intraclass Correlation Coefficient (ICC) was used to evaluate intraand inter-observer agreement. To evaluat e t h e G a ussian distribution of the data, the Shapiro-Wilk test was performed and the MannWhitney test was applied for independent variables, between sexes, for measurements of the FS region. Regarding the facial skeletal pattern, the analysis of variance (ANOVA) was applied to compare the different FS analyses. For the age, Kruskal-Wallis was applied to compare the different FS analyses. To develop a formula capable of estimating sex through FS measurements, a multivariate discriminant analysis was carried out using the direct method, and for this purpose, the Wilks’ lambda and Box’s M tests were applied. 26 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 RESULTS In table 3, it is possible to observe that the volume, width and depth of the FS showed significant differences between female and male. Female individuals exhibited significantly smaller volume, width and depth than men (pvalue<0.001, 0.003 and <0.001, respectively). In tables 4 and 5, no statistically significant differences were found between the groups separated by age and facial skeletal pattern. Table 3. FS parameters according to sex Mann-Whitney test; SD (standard deviation); *p<0.05 Table 4. FS parameters according to sex Sex N Mean Median SD p-value Volume (mm3) F 164 4579.76 3966.00 2739.13 <0.001* M 71 7288.35 6344.00 4366.16 Height (mm) F 164 24.03 25.05 8.49 0.958 M 71 24.26 25.50 7.74 Width (mm) F 164 38.04 40.55 11.55 0.003* M 71 42.87 45.70 13.39 Depth (mm) F 164 7.34 7.30 2.19 <0.001* M 71 9.37 9.10 2.49 Group N Mean Median SD p-value Volume 20-24.9 60 5662.03 5335 475.684 0.240 25-29.9 50 5419.88 4262 503.215 30-34.9 20 6459.90 5509 939.075 35-39.9 26 6058.31 4302 842.146 40-45.9 19 5405.37 6070 690.354 46-49.9 25 4165.20 3769 543.905 50-54.9 16 3715.06 2823 783.026 55-59.9 10 5064.11 5056 785.282 60-64.9 9 5601.33 5433 712.646 Height 20-24.9 60 24.53 26.8 1.043 0.127 25-29.9 50 23.20 24.5 1.159 30-34.9 20 25.93 25.9 1.843 35-39.9 26 25.97 26.2 1.502 40-45.9 19 25.95 26.4 2.157 46-49.9 25 21.83 23.7 1.796 50-54.9 16 19.49 19.0 2.334 55-59.9 10 23.52 23.8 1.718 60-64.9 9 27.04 26.3 1.288 Width 20-24.9 60 39.62 40.5 1.600 0.119 25-29.9 50 39.61 41.1 1.581 30-34.9 20 43.79 42.5 2.516 35-39.9 26 40.57 41.3 2.365 40-45.9 19 40.18 43.3 3.288 27 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Kruskal-Wallis test; SD (standard deviation); *p<0.05 Table 5. FS parameters according to facial skeletal pattern ANOVA test; SD (standard deviation); *p<0.05 Ta bl e 6 sho ws th e Wilks ' La mbd a t est , app li ed t o evaluate the ability of the discriminant equation to distinguish the sexes based on the values of the FS volume, height, width and depth analyses. The Wilks' Lambda value was 0.834, with a p-value of <0.001, indicating that the discriminant equation is statistically significant and has discriminatory power to distinguish the sexes. Furthermore, it is possible to observe that the overall percentage of precision in predicting gender was 74.5%, which indicates a moderate capacity for accuracy in identifying gender based on the measurements considered. The function values in the centroid group were 0.675 for males and -0.292 for females. A rule was established indicating that if the D value is greater than 0.19, the sample will be classified as male. These results suggest that FS volume, height, width and depth measurements have moderate discriminatory power to predict sex in a Brazilian population. 46-49.9 25 35.61 40.9 2.668 50-54.9 16 32.61 34.0 3.536 55-59.9 10 40.48 43.3 3.711 60-64.9 9 44.20 43.4 1.843 Depth 20-24.9 60 8.15 8.40 0.333 0.983 25-29.9 50 8.00 8.15 0.386 30-34.9 20 7.99 8.40 0.482 35-39.9 26 8.07 7.45 0.491 40-45.9 19 7.51 8.30 0.547 46-49.9 25 7.80 7.80 0.498 50-54.9 16 7.53 8.25 0.632 55-59.9 10 8.12 8.80 0.695 60-64.9 9 7.67 7.20 0.480 Class N Mean SD p-value Volume (mm3) I 89 5289.67 3176.82 0.932 II 83 5443.00 3567.69 III 63 5492.13 3990.49 Height (mm) I 89 23.94 7.56 0.073 II 83 25.54 8.70 III 63 22.40 8.38 Width (mm) I 89 39.70 12.06 0.980 II 83 39.35 12.13 III 63 39.40 13.06 Depth (mm) I 89 7.91 2.50 0.084 II 83 8.38 2.38 III 63 7.47 2.46 28 JFOS - Journal of Forensic Odonto-Stomatology Vol 43 n. 2 - Aug - 2025 Table 6. Discriminant analysis of sex, using the measures in discrimination DISCUSSION To our knowledge, this is the first study to discuss morphometric measurements of FS in relation to age, sex and skeletal class in a Brazilian population. We observed that the FS showed significant differences between the sexes, with females exhibiting smaller volume, width and depth compared to males (p-value<0.001, 0.003 and <0.001, respectively). However, it was not possible to observe statistically significant differences in relation to age and skeletal class. Although some studies have explored the volume of FS in assessing sex, its practical application in forensic anthropology is still in its infancy.1 In the current scientific literature, based on helical computed tomography (CT) or CBCT, the significant distinction in the morphological characteristics of SF in relation to sex is reinforced.1,4,12,13,28,29 Nevertheless, there are works that question the reliability of FS as a forensic identification method,14-17 many of which employ 2D examinations and in some cases, resort to the use of dry skulls. In this present approach, using CBCT, the limitations of twodimensional images can be surpassed, considering that 3D images provide details without superimposition, improving analysis and measurements, allowing more accurate results.11 Regarding age, our results did not indicate statistically significant correlations, suggesting a complex relationship between FS and aging. Even so, there are some studies that have shown growth of the FS until the age of 40, and its morphology can be influenced due to factors such as mechanical stress and growth hormones.18,19 Our sample, with n=9 for the group over 60 years old, did not reflect this correlation. This points to the need for more comprehensive investigations involving different age groups. Exploring the correlation between FS and facial skeletal pattern, the results of this research did not reveal statistically significant associations, similar to the pilot study by Sawada et al.9 However, it differed from two previous studies.7,8 Metin-Gürsoy et al. (2021) when performing linear measurements on CBCT with individuals aged 17 to 38 years, found that the anteroposterior dimension and width were significantly smaller in the hyperdivergent group compared to the hypodivergent, revealing a correlation between FS and craniofacial measurements. Rossouw et al. (1991) analyzed the FS using 2D cephalometry and concluded that there is a significant relationship between this structure and skeletal class. Therefore, our results suggest that FS may not be a reliable indicator for facial skeletal pattern, highlighting the complexity of these relationships and the need for additional investigations. It is worth highlighting that variations in the size and capacity of the FS may be related to its development during puberty, in parallel with craniofacial growth.7 The present study has some limitations, such as the convenience sample and individuals aged at least 20 years.1 This age is considered the end of the FS growth. For this reason, measurements and comparisons were not considered in people of younger age.1 Expanding the sample is necessary to better represent a population.1 The present sample showed an inequality between the facial skeletal pattern and age groups, therefore requiring the expansion of the sample, following standardized protocols for different populations. However, the search for a reliable, accessible and accurate method for human identification motivates the continuation of these studies, providing valuable information for forensic science. Total (sex) D = -1.125 + 0.00*volume -0.005*height - 0.001*width + 0.12*depth Wilks’ Lambda = 0.834, p<0.001 Percentage of accuracy in predicting sex Overall 74.5% Functions at centroid group Male Female Male 0.675 -0.292 if D > 0.19 29