Gender-Based Dimorphism of Maxillary and Sphenoid Air Sinuses Via 3D Volumetric Segmentation of CBCT in a Sample of Egyptians

By Ehab Samir Ali Abuelola, Ola Mohamed Rehan, Hany Omar

Affiliations

1. Department of Oral and Maxillofacial Radiology, Faculty of Dentistry, Cairo University, Cairo, Egypt

doi: 10.29271/jcpsp.2025.07.843

ABSTRACT
Objective: To determine gender in forensic odontology by volume calculation of maxillary and sphenoid sinuses via three-dimensional (3D) semi-automatic segmentation using cone beam computed tomography (CBCT) datasets in a sample of Egyptians.
Study Design: A cross-sectional descriptive study.
Place and Duration of the Study: Department of Oral and Maxillofacial Radiology, Faculty of Dentistry, Cairo University, Cairo, Egypt, from January to December 2024.
Methodology: The study included 204 CBCT scans (102 males and 102 females) collected from the database of the study centre. Patients were divided into three age ranges: Group I (18-23 years), Group II (24-28 years), and Group III (>28 years). The maxillary sinus (MS) and sphenoid sinuses (SS) were segmented by MIMICS software. Associations with gender and age were analysed using the one-way ANOVA test. Correlation analysis was made using Spearman's rank-order correlation coefficient. Linear discriminant analysis (LDA) was then implemented for gender discrimination.
Results: Maxillary sinus volume (MSV) and sphenoid sinus volume (SSV) were higher in males (p <0.05). For gender determination, the most discriminant parameter was left MSV, with an overall accuracy of 63.24%, that decreased to 62.25% when all evaluated sinuses were combined. No significant difference was detected in sinus volumes among different age groups. For age estimation, the right MSV and the average volume of all sinuses showed low overall accuracy, both at 36.80% which increased to 37.70% when all evaluated sinuses were combined.
Conclusion: Sphenoid and maxillary air sinuses may be used as valid anatomical landmarks for gender determination. More studies including larger sample sizes of different populations are necessary.

Key Words: Forensic imaging, CBCT, Sphenoid air sinus, Maxillary air sinus, Gender discrimination.

INTRODUCTION

Since ancient times, humankind has tried to identify people, animals, and objects to resolve unclear situations. Legal and social needs drive the necessity for human identification. Recognising someone's identity, whether during their life or post- mortem, ensures full acknowledgement of their rights and duties.¹ Recognising a biological profile necessitates four pieces of information: A person's height, gender, age at the death time, and ethnic descent.¹ Gender determination is essential in constructing an individual’s biological identifying characteristics as it affects both height and estimation of age at the death time.²

Traditionally, bones used for identification, such as long bones, skull bones, and pelvic bones are frequently found mutilated. Therefore, it has become crucial to rely on harder bones that may be recovered undamaged. Paranasal sinuses maintain their integrity even in extreme conditions and can withstand adverse environmental factors due to their distinct characteristics.³  So far, the most reliable methods of identification have included fingerprints, dental compari- son, and biological techniques such as deoxyribonucleic acid (DNA) profiling.⁴ However, DNA is often demolished when remains are incinerated and decomposed. Furthermore, in the absence of prior dental record, these identification methods cannot be utilised.^5,6

 Consequently, maxillofacial imaging technologies play a vital role in forensic anthropology by evaluating the morphology and dimensions of the sinuses. Three-dimensional (3D) structure analyses provide more detailed information than conventional two-dimensional (2D) imaging. Cone beam computed tomography (CBCT) offers 3D multi-planar images with low doses of radiation, less capturing time, and reduced costs compared to conventional CT.⁷

Most studies concerted on the maxillary sinus (MS) linear measurements. However, a valid protocol for 3D volume assessment is still required. Forensic investigations would benefit from more precise protocols using CBCT software to reconstruct 3D virtual models. Measurement of volume using programmes that allow segmentation and modelling are compatible with 3D-visualisation methods and allow a morphometric assessment. Therefore, comparing maxillary sinus volume (MSV) and sphenoid sinus volume (SSV) based on gender is deemed valuable.⁸

A systematic review conducted by Sidhar et al. on 41 articles of mainly Indian and Iranian populations and to a smaller number of Turkish, Egyptian, and other populations concluded that more studies of large samples using more standardised protocols of measurements are mandatory.⁹ Different racial populations also have to be studied to assess the variety and reliability of these measurements. Consequently, this study aimed to assess the accuracy of 3D volumetric segmentation using CBCT datasets in the gender dimorphism of MS and SS in a sample of Egyptians.

METHODOLOGY

This study was approved by the Ethics Committee, Faculty of Dentistry, Cairo University (code no. 6623). A total of 204 CBCT scans belonging to Egyptian patients were grouped equally into males and females (102 patients in each group) and anonymised by software. The patients were collected from the database of the Department of Oral and Maxillofacial Radiology, Faculty of Dentistry, Cairo University, Cairo, Egypt. Scans were obtained using the CBCT scanner Planmeca Promax imaging system (Planmeca Oy, Helsinki, Finland), with exposure parameters adjusted according to patient size and manufacturer’s recommendations. According to age range, patients were segmented into three groups: Group I (18-23 years), Group II (24-28 years), and Group III (>28 years), with each group comprising 68 patients. Inclusion criteria included adult Egyptian patients of both genders ≥18 years, a complete dentate maxillary dental arch-with or without wisdom teeth, and normal anatomical representation of examined air sinuses. Scans with sinus lesions or thickening of its mucous lining, facial pathological lesions, skeletal asymmetries, craniofacial trauma, fractures, or developmental abnormalities, as well as scans with inadequate fields of view, suboptimal resolution, or significant artefacts were excluded.

CBCT digital imaging and communication in medicine (DICOM) files were imported into MIMICS software (version 21.0; Materialise, Leuven, Belgium). A radiologist with 10 years of experience conducted the segmentation process. To ensure inter-rater consistency, fifty cases were re-segmented by another radiologist with 15 years of experience. Semi-automatic segmentation of right and left MS and SS was performed using the dynamic region-growing tool. To ensure segmentation accuracy, thresholding was then applied to the resulting mask to remove non-sinusoidal data. The volumetric filling was applied with a preselected colour, and the volume of the final 3D sinus model was calculated in cubic millimetres (mm³) by the software and converted to cubic centimetres (cm³) for better visualisation of results (Figure 1).

Figure 1: Volumetric filling of (A) maxillary (green) and (B) sphenoid (blue) sinuses. (C) 3D volumetric segmentation of maxillary (green) and sphenoid sinuses (blue) using MIMICS software.

Statistical analysis was performed with R statistical analysis software version 4.4.1 for Windows.

Categorical data were represented as frequency (percentages). Continuous data were presented as mean and standard deviation (SD). Normality was tested by Shapiro-Wilk’s test and the data revealed normal distribution. Associations with gender and age were analysed using the one-way ANOVA test. Spearman's rank correlation coefficient was used for correlation analysis. LDA was applied to assess the ability of different variables to differentiate gender and age. Homogeneity of variances was checked using the Box M test, and for multivariate models, multicollinearity was evaluated by the variance inflation factor (VIF), and variables recorded VIF values ​​>5 were excluded from the models. The significance level was set at p <0.05.

RESULTS

The mean right MSV was 15.74 ± 5.17 cm³. For left MSV, the mean value was 15.77 ± 5.48 cm³. For the entire MSV, the mean value was 15.75 ± 5.21 cm³. For SSV, the mean value was 10.19 ± 4.35 cm³. For all sinuses combined, the mean value was 12.97 ± 4.30 cm³.

Right and left MSV and SSV of males were obviously higher than females (p <0.05) as shown in Table I. No significance was revealed in sinus volumes across different age groups.

Strong positive statistically significant correlations (rs >0.5, p <0.001) were observed between the right and left MSV (rs = 0.927), left MSV and SSV (rs = 0.611), and right MS and SS (rs = 0.595).

Table I: Mean, standard deviation, and p-value of right and left MSV and SSV in both genders.

Measurements (cm3)	Mean ± SD		f-value	p-value
	Male	Female
Right MSV	17.17 ± 4.99	14.31 ± 4.97	16.87	<0.001*
Left MSV	17.13 ± 5.39	14.40 ± 5.24	13.41	<0.001*
MSV (average)	17.15 ± 5.06	14.36 ± 5.00	15.74	<0.001*
SSV	11.08 ± 4.40	9.31 ± 4.14	8.69	0.004*
Average sinuses	14.11 ± 4.19	11.83 ± 4.12	15.35	<0.001*
*Significant. ANOVA test significance level <0.05; MSV, Maxillary sinus volume; SSV, Sphenoidal sinus volume.

Table II: Univariable models for gender determination.

Measurements	Coefficient	Fisher’s linear DF		Centroids		Accuracy of cross-validated predictions (%)
		Male	Female	Male	Female	Sensitivity Male prediction	Specificity Female prediction	Overall Accuracy
Constant	-3.16	-6.64	-4.82	0.288	-0.288	59.34%	63.73%	58.33%
Right maxillary sinus	0.20	0.69	0.58
Constant	-2.97	-5.89	-4.36	0.256	-0.256	64.52%	67.65%	63.24%
Left maxillary sinus	0.19	0.61	0.51
Constant	-3.13	-6.50	-4.76	0.278	-0.278	63.04%	66.67%	61.76%
Right and left maxillary sinuses (average)	0.20	0.68	0.57
Constant	-2.39	-4.05	-3.07	0.206	-0.206	59.80%	59.26%	56.86%
Sphenoid sinus	0.23	0.61	0.51
Constant	-3.12	-6.46	-4.75	0.274	-0.274	61.17%	60.78%	61.27%
All sinuses (average)	0.24	0.82	0.69

Table III: Univariable models for age estimation.

Measurements	Coefficient	Fisher’s linear DF			Centroids			Accuracy of cross-validated predictions (%)
		18-23	24-28	>28	18-23	24-28	>28
Constant	-3.05	-5.31	-6.16	-5.81	-0.150	0.131	0.019	36.80%
Right maxillary sinus	0.00	0.001	0.001	0.001
Constant	-2.88	-4.95	-5.58	-5.12	-0.103	0.117	-0.013	34.30%
Left maxillary sinus	0.00	0.001	0.001	0.001
Constant	-3.03	-5.29	-6.06	-5.68	-0.128	0.126	0.002	34.80%
Right and left maxillary sinuses (average)	0.00	0.001	0.001	0.001
Constant	-2.35	-3.71	-4.32	-3.59	-0.068	0.186	-0.118	36.30%
Sphenoid sinus	0.00	0.001	0.001	0.001
Constant	-3.12	-5.50	-6.36	-5.75	-0.121	0.157	-0.036	36.80%
All sinuses (average)	0.24	0.001	0.001	0.001

Analysis of linear discriminant function (LDA) to determine gender was employed. Univariable models revealed that left MSV was the most accurate discriminant parameter with an overall accuracy of 63.24%, sensitivity of 64.52% (accurate male predictions), and specificity of 67.65% (accurate female predictions) as shown in Table II. The multivariable model, which used the average of right and left MSV as well as the SSV, showed a lower overall accuracy of 62.25%, with 63.44% for males and 66.67% for females.

For age estimation, LDA also yielded low values. Univariable models indicated that the right MSV and the average volume of all sinuses had the highest overall accuracy, both of 36.80% as shown in Table III. The multivariable model showed an overall accuracy of 37.70%. Inter-observer reliability revealed good agreement (ICC = 0.899) of the right MS measurements and excellent agreement of the left MS and SS measurements (ICC >0.9).

DISCUSSION

To enhance the role of radiography in the identification process of individuals, anatomical measurements, visual examination, and exact measurement of bone dimensions are performed, particularly when skeletal remains are involved. Cranium is very helpful in radiographical comparisons, as it contains the paranasal sinuses which typically keep its integrity even when severe damage occurs to other bones. SS is of the prime concern because of its deep anatomical location, making it less prone to damage and pathological conditions.¹⁰ Meanwhile, MS, which has the largest volume among all the paranasal air sinuses, plays a critical role in the development of face morphology.

Many previous studies revealed that MSV is greater in males.^8,11-14 Also, SSV in adult patients are significantly greater in males than in females, which may help in gender dimor-phism.^1,14-16 Thus, in this study the volume of these sinuses in adult patients was examined if it can be used in gender discrimination. By the age of 12 years, the MS's width and length have grown to adult proportions, and its height keeps rising until the age of 18 years.¹⁷ The SS reaches its adult size around 12 years of age, with some authors stating it reaches adult size at 18 years.¹⁶ Thus, patients included in the current study were aged ≥18 years.

Only patients of full dentate maxillary arches were included in this study, as the dentition status of the individuals and loss of maxillary posterior teeth may lead to obvious alteration in MSV due to the resumption of the pneumatisation process after teeth extraction.⁷ Likewise, there is a relation between the dental condition and SSV due to the masticatory force applied from the teeth to the cranium. Thus, to avoid possible changes in the measurements taken in this study, images of only patients with complete dentition were used.¹

By removing superimposition and limiting magnification, CBCT allows precise measurements and increases the relevance of 3D imaging for morphometric investigation of cranium in dental forensic science.¹⁸ Since sinuses do not have uniform borders, linear measurements may differ from reality; therefore, segmentation methods are considered more precise and reliable than linear measurements.⁸ There are many segmentation methods: Manual, semi-automatic, and totally automatic. In the manual method, the area of interest is defined slice by slice, and then all successive slices are combined to form the whole volume, making it very time- consuming and operator-dependent. On the other hand, automatic segmentation may be associated with inaccuracies in complex structures, which is why semi-automatic methods are desired.¹⁹ Thus, in the current study, 3D segmentation using a semi-automatic method was achieved by combining the automatic and manual procedures using MIMICS software to measure MSV and SSV.

In the current study, MSV was larger in males than females (p <0.001). These findings align with those of Radulesco et al.,¹³ Aktuna Belgin et al.,⁸ and Wanzeler et al.’s findings.¹⁴ Additionally, Asantogrol et al. found that the right MSV was significantly higher in males than females.²⁰ These findings may be attributed to differences in the overall size of the skull between both the genders. Other studies could not find statistically significant differences.^21,22 Additionally, this study revealed that the SSV was significantly higher in males (p = 0.004). This result is equivalent to the findings of Gibelli et al.,¹⁵ Wanzeler et al.,¹⁴ Ramos et al.,¹ and Banihashem Rad et al.¹⁶ but differs from Oliveira et al.,²³ who found no significant differences between gender and sinus volume. The variation in results can be attributed to factors such as differences in gender distribution, racial backgrounds, and the various methods used to measure the volume of sinuses.

The results of this study revealed that the left MSV was the parameter with the highest accuracy of 63.24%. While multivariable model using the average of right and left MSV as well as SSV had a lower overall accuracy of 62.25%. Wanzeler et al. found that gender identification using both the right and left MSV was 83.7% in males, and 85.5% in females, and the overall accuracy was 84.66%.¹⁴ When all sinuses volumes were included, the accuracy increased to 96.2% in males and 92.7% in females, and the overall accuracy was 94.48%. Banihashem Rad et al. revealed that the accuracy of gender differentiation using SSV was 86.0% in males and 92.9% in females, while the overall accuracy was 90.2%.¹⁶ The variations in these findings can be attributed to variances in population, age range, and different software programmes utilised in volume calculation.²⁴

In this study, subjects were segmented into three age range groups with 5-year intervals to ensure the occurrence of considerable changes in the sinuses volume.²⁵ No significant difference was detected in sinus volumes among these age groups. For age estimation, the results of this study revealed low accuracy, with the right MS and the average volume of all sinuses having the highest overall accuracy, both at 36.80%. The multivariable model showed an overall accuracy of 37.70%. Likewise, Radulesco et al.¹³ and Gulec et al.²¹ reported no correlation between age and MSV. Also, Oliveira et al.²³ noted no correlation between SSV and age. In contrast, Aktuna Belgin et al.⁸ reported significant variations in MSV among different age groups. Diverse results can be explained by differences in age categorisation. To current knowledge, there is a deficiency in studies on age prediction using 3D volumetric segmentation, indicating a need for further research.

CONCLUSION

MSV and SSV may be useful for gender determination, as there was a significant difference in sinus volumes between females and males. The most discriminant parameter for gender determination was the left MSV. However, age estimation revealed poor outcomes, with no apparent difference between the sinuses volume and age. More studies incorporating large sample sizes are needed to support such results. Additionally, since this study was retrospective, it was not possible to examine the correlation between the measurements of MS and SS and the patients’ size or body stature.

ETHICAL  APPROVAL:
Ethics committee, Faculty of Dentistry, Cairo University approved this study (code number: 6623).

PATIENTS’  CONSENT:
Informed consent was waived by the Ethics Committee, Faculty of Dentistry, Cairo University, Cairo, Egypt.

COMPETING  INTEREST:
The authors declared no conflict of interest.

AUTHORS’  CONTRIBUTION:
ESAA: Conception and design of the work, acquisition, analysis, interpretation of data, and drafting of the work.
OMR: Acquisition, data analysis, and revising the manuscript critically for important intellectual content.
HO: Final approval of the version to be published. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy and integrity of any part of the work are appropriately investigated and resolved.
All authors approved the final version of the manuscript to be published.

REFERENCES

1. Ramos BC, Manzi FR, Vespasiano AI. Volumetric and linear evaluation of the sphenoidal sinus of a Brazilian population, in cone beam computed tomography. J Forensic Leg Med 2021; 77:102097. doi: 10.1016/j.jflm.2020.102097.
2. Peckmann TR, Orr K, Meek S, Manolis SK. Sex determination from the talus in a contemporary Greek population using discriminant function analysis. J Forensic Leg Med 2015; 33:14-9. doi: 10.1016/j.jflm.2015.03.011.
3. Sidhu R, Chandra S, Devi P, Taneja N, Sah k, Kaur N. Forensic importance of maxillary sinus in gender determination: A morphometric analysis from Western Uttar Pradesh, India. Eur J Gen Dent 2014; 3(1):53-6. doi: 10.4103/2278-9626.126213.
4. Hemanthakumar S, Gopal KS, Kumar PM. Assessment of sexual dimorphism using3D CBCT image data among Indians. Bioinformation 2022; 18(3):231-8. doi: 10.6026/97320630018231.
5. Soares CBRB, Miranda-Viana M, Pontual AA, Ramos-Perez FMM, Perez DEC, Figueiroa JN, et al. Morphological and dimensional assessment of the maxillary sinus for human identification and sexual dimorphism: A study using CBCT. Forensic Imaging 2020; 23:200409. doi: 10.1016/j.fri.2020.200409.
6. Patil K, Mahesh KP, Sanjay CJ, Vijayan MA, Nagabhushana D, Ramesh AA. Bizygomatic distance as a predictor of age and sex determination: A morphometric analysis using cone beam computed tomography. Eur J Anat 2022; 26(4):457-63. doi: 10.52083/WGES2700.
7. Deshpande AA, Munde AD, Mishra SS, Kawsankar KD, Sawade RV, Mandar B. Determination of sexual dimorphism of the 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. doi: 10.4103/jfmpc.jfmpc_389_21.
8. Aktuna Belgin C, Colak M, Adiguzel O, Akkus Z, Orhan K. Three- dimensional evaluation of maxillary sinus volume in different age and sex groups using CBCT. Eur Arch Otorhinolaryngol 2019; 276(5):1493-9. doi: 10.1007/s00405-019-05383-y.
9. Sidhar M, Bagewadi A, Lagali-Jirge V, S LK, Panwar A, Keluskar V. Reliability of gender determination from paranasal sinuses and its application in forensic identification-a systematic review and meta-analysis. Forensic Sci Med Pathol 2023; 19(3):409-39. doi: 10.1007/s12024-022-00520-2.
10. Mohebbi A, Rajaeih S, Safdarian M, Omidian P. The sphenoid sinus, foramen rotundum and vidian canal: A radiological study of anatomical relationships. Braz J Otorhinolaryngol 2017; 83(4):381-7. doi: 10.1016/j.bjorl.2016.04.013.
11. Demir UL, Akca ME, Ozpar R, Albayrak C, Hakyemez B. Anatomical correlation between existence of concha bullosa and maxillary sinus volume. Surg Radiol Anat 2015; 37(9):1093-8. doi: 10.1007/s00276-015-1459-y.
12. Prabhat M, Rai S, Kaur M, Prabhat K, Bhatnagar P, Panjwani S. Computed tomography based forensic gender determination by measuring the size and volume of the maxillary sinuses. J Forensic Dent Sci 2016; 8(1):40-6. doi: 10.4103/0975- 1475.176950.
13. Radulesco T, Michel J, Mancini J, Dessi P, Adalian P. Sex estimation from human cranium: Forensic and anthropological interest of maxillary sinus volumes. J Forensic Sci 2018; 63(3):805-8. doi: 10.1111/1556-4029.13629.
14. Wanzeler AM V, Alves-Junior SM, Ayres L, da Costa Prestes MC, Gomes JT, Tuji FM. Sex estimation using paranasal sinus discriminant analysis: A new approach via cone beam computerized tomography volume analysis. Int J Legal Med 2019; 133(6):1977-84. doi: 10.1007/s00414-019-02100-6.
15. Gibelli D, Cellina M, Gibelli S, Oliva AG, Codari M, Termine G, et al. Volumetric assessment of sphenoid sinuses through segmentation on CT scan. Surg Radiol Anat 2018; 40(2): 193-8. doi: 10.1007/s00276-017-1949-1.
16. 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 (Shiraz) 2023; 24(1 Suppl):95-102. doi: 10.30476/dentjods.2022.92915.1685.
17. Whyte A, Boeddinghaus R. The maxillary sinus: Physiology, development and imaging anatomy. Dentomaxillofac Radiol 2019; 48(8):20190205. doi: 10.1259/dmfr.20190205.
18. Ahmed J, Namrata, Sujir N, Shenoy N, Natarajan S, Muralidharan A, et al. A comparative analysis of sphenoid and frontal sinuses using cone beam computed tomography for sex determination. J Oral Biol Craniofac Res 2024; 14(4):478-83. doi: 10.1016/j.jobcr.2024.05.004.
19. Vallaeys K, Kacem A, Legoux H, Le Tenier M, Hamitouche C, Arbab-Chirani R. 3D dento-maxillary osteolytic lesion and active contour segmentation pilot study in CBCT: Semi-automatic vs. manual methods. Dentomaxillofac Radiol 2015; 44(8):20150079. doi: 10.1259/dmfr.20150079.
20. Asantogrol F, Etoz M, Topsakal KG, Can FE. Evaluation of the maxillary sinus volume and dimensions in different skeletal classes using cone beam computed tomography. Ann Med Res 2021; 28(4):709-15. doi: 10.5455/annalsmedres.2020.05.529.
21. Gulec M, Tassoker M, Magat G, Lale B, Ozcan S, Orhan K. Three- dimensional volumetric analysis of the maxillary sinus: A cone-beam computed tomography study. Folia Morphol (Warsz) 2020; 79(3):557-62. doi: 10.5603/FM.a2019.0106.
22. Sarilita E, Lita YA, Nugraha HG, Murniati N, Yusuf HY. Volumetric growth analysis of maxillary sinus using computed tomography scan segmentation: A pilot study of Indonesian population. Anat Cell Biol 2021; 54(4):431-5. doi: 10.5115/ acb.21.051.
23. Oliveira JM, Alonso MB, de Sousa E Tucunduva MJ, Fuziy A, Scocate AC, et al. Volumetric study of sphenoid sinuses: Anatomical analysis in helical computed tomography. Surg Radiol Anat 2017; 39(4):367-74. doi: 10.1007/s00276-016- 1743-5.
24. An G, Hong L, Zhou XB, Yang Q, Li MQ, Tang XY. Accuracy and efficiency of computer-aided anatomical analysis using 3D visualization software based on semi-automated and automated segmentations. Ann Anat 2017; 210:76-83. doi: 10.1016/j.aanat.2016.11.009.
25. Heo SJ, Jee YS. Characteristics of age classification into five-year intervals to explain sarcopenia and immune cells in older adults. Medicina (Kaunas) 2023; 59(10):1700. doi: 10.3390/ medicina59101700.