INTRODUCTION

Sex determination of a body is a primary forensic procedure in the identification process. If primary methods of identification are impossible (absence of comparative DNA samples, fingerprints, or teeth records), it can be performed using radiographic techniques, whenever ante mortem images are available (Lorkiewicz-Muszynska et al., 2009). In violent traumatic contexts, teeth and frontal sinuses are likely to be missing, or separate into small pieces from the rest of the body, which makes personal identification unable to proceed. Sphenoid sinus, on the other hand, remains to be the most hidden and unreachable of the paranasal sinuses (^{Mamatha et al., 2010}; ^{Sevinc et al., 2014}; ^{Wiebracht et al., 2014}). It continues to develop during puberty and reaches adult size by the age of 12 (Wiebracht et al., 2014). Deeply positioned in the center of the cranial base, it is well protected from degradation resulting from external force (^{Casselman et al., 2013}).

It was reported that anatomy of sphenoid sinus could be used in personal identification, but to our knowledge, the significance of both qualitative and quantitative indicators of sphenoid sinus in the field of forensic medicine are rarely assessed. It is comprehensible that traditional cranial radiographies can not visualize them precisely due to the crucial bone structure superimposition (^{Auffret et al., 2016}). Although CT can reflect the bony structure well, the sphenoid sinus is connected to the sieve like ethmoid sinus, which makes the single three-dimensional reconstruction of sphenoid sinus difficult to measure. Not using three-dimensional reconstruction alone for the measurement of sphenoid sinus, MRI can depict the true anatomy of deep tissue more clearly without bone structure superimposition. It can be used to determine the anatomic structure of the lesion region in detail. More importantly, it can be used to measure the sphenoid sinus at the soft tissue level around it. The aim of this study was therefore to evaluate whether sexual dimorphism from the height, A-P diameter, area and perimeter of sphenoid sinuses could be determined using the MRI imaging modality.

MATERIAL AND METHOD

Materials and instruments. Nuclear magnetic resonance machine GE (Discoverery, MR750w 3.0T, America); Mimics Research software (version 19.0, Materialise Inc,Belgium); Photoshop image processing software; SPSS20.0 statistics software.

Selection of subjects. The present study was carried out in the Department of Radiology, affiliated hospital of Shanxi Medical University, after required approval from the Institutional Ethical Committee. A prospective study comprising 79 subjects (44 males and 35 females) ranged from 27 to 87 years age group over a period of 13 months (June 2018 to July 2019) was selected for the purpose of this study. Healthy subjects were randomly selected who attends the outpatient department and informed consent was obtained. Subjects with the medical history of facial trau- ma, tumors or any pathological process concerning the sphenoid bone and the other surrounding cranial base bones, but also with sinuses mucosa thickening, or any abnormality of the sinuses contents, were excluded from the study.

Data acquisition. All subjects were examined on MRI. All the scans were made using a nuclear magnetic resonance machine GE (Discovery, MR750w 3.0T, America). All images were converted to Digital Image Communication in Medicine (DICOM) format. The observer was blind to the details of age and sex of the subjects, using the DICOM compatible Mimics Research software (version 19.0, Materialise Inc, Belgium) to measure the height, A-P diameter, area and perimeter in the midsagittal view of the sphenoid sinus.

The following measurements were performed:

The height was measured as the vertical distance from the lowest point to the highest point in the midsagittal view (Fig. 1).

The anterior-posterior diameter was defined as the perpendicular distance from the most anterior point to the most posterior point in the midsagittal view (Fig. 2).

The area and perimeter were automatically estimated by drawing the outline of sphenoid sinus in the midsagittal view (Fig. 3).

Analysis of data. All data were subjected to descriptive analysis where comparison between both sexes was done using unpaired t test with a p value less than 0.05 taken as significant level. Canonical discriminant with sex as a grouping variable and sphenoid sinus measurements as independent variables was performed to generate a Canonical discriminant function. Canonical discriminant function provided by the model calculates the discriminate score, which can be used to predict sex by substituting the value of the specific measurements into the equation. The discriminate score is compared with reference values (also provided by the model): functions at group centroids (one for female, the other for male). If the discriminate score is closer to one of the reference values, it indicates we can predict sex accordingly. The predictive accuracy rate of the results and reliability of the procedure was assessed. Data analysis was done using the software, SPSS version 20.0.

Fig. 1 The height in the midsagittal view of the sphenoid sinuse. 

Fig. 2 The A-P diameter in the midsagittal view of the sphenoid sinuse. 

Fig. 3 The area and perimeter in the midsagittal view of the sphenoid sinuse. 

RESULTS

In the current study, all the parameters of sphenoid sinuses of the male were found to be greater than those of female. Table I shows the descriptive analysis of the measurements of sphenoid sinuses according to sex where the mean height, A-P diameter, area and perimeter with their standard deviations (SD) were presented. Significant mean differences were observed among all the four parameters between both sexes (Figs. 4-7).

Canonical discriminant function was obtained as follows: sex=- 3,419+0.007·the area in the midsagittal view of the sphenoid sinuses. The obtained determinant equation was applied to the study sample and revealed that the predictive accuracy rate of the sphenoid sinus to identify sex was 63.6 % in males and 62.9 % in females with an overall accuracy of 63.3 % (Table II).

Fig. 4 Comparison of sphenoid sinus height between sexes in histogram. 

Fig. 5 Comparison of sphenoid sinus A-P diameter between sexes in histogram. 

Fig. 6 Comparison of sphenoid sinus area between sexes in histogram. 

Fig. 7 Comparison of sphenoid sinus perimeter between sexes in histogram. 

Table I Comparison of parameters among male and female groups of sphenoid sinuses. 

Table II Classification results of discriminant functional analysis of sphenoid sinus. 

DISCUSSION

In the field of forensic science, sex determination of a body is a crucial initial step in the identification process. Currently there are many techniques for sex identification in forensic science, the most reliable can being DNA analysis. However, this approach does not work in the event of a complete loss of soft tissue due to a major disaster such as fire, tsunami, or plane crash. In such cases, other methods can be used like radiographic techniques, whenever ante mortem images are available. MRI, being a three-dimensional radiographic technique, is radiation-free and without bone superimposition and thus allows a better assessment of the soft tissue and deep cranial tissue such as the sphenoid sinus.

It was reported that the accuracy rate of sex determination is 100 % from a complete skeleton, 98 % from both the pelvis and the cranium, 95 % from the pel- vis and the long bones, and 80-90 % from the long bones only (^{Teke et al., 2007}). In cases of mass disasters, even the skull and other bones are severely blemished, it has been reported that paranasal sinuses remain intact because of their gas-filled bone structures. Among the four pairs of paranasal sinuses, the sphenoid sinus is the deepest and the least susceptible to injury.

^{Zecchi et al. (1983}) studied the three dimensions and the volume of the sphenoid sinus of 300 human cranial radiograms and stated that the measurements and volume of sphenoid sinus between both sexes were significantly different. A study conducted by ^{Oliveira et al. (2009}) about sex determination from area and volume of the sphenoid sinuses showed that the sphenoid sinus was larger in males than in females and thus can be used to evaluate sexual dimorphism.

It was described that there is a significant variation in the structure of the sphenoid sinus, which contributes to the few papers published defining the sphenoidal configuration and its anatomy (^{Fujioka & Young, 1978}). In 2014, ^{Yamashita et al. (2014}) divided the sphenoid sinus variations into four types as conchal, presellar, postsellar and sellar. In 2016, ^{Auffret et al. (2016}) established an anatomical classification system of the sphenoid sinuses anatomical variations and evaluated the value of visual comparison of the CT anatomical aspects of the sphenoid sinuses, and indicated that visual comparison of sphenoid sinuses CT could be an efficient, non-invasive method for personal identification.

The results of present study showed that the height, anterior-posterior diameter, area, and perimeter in the midsagittal view of the sphenoid sinuses was lager in ma- les as compared to females which is similar to the above research.

However, our study has limitations, such as selecting only living volunteers, while postmortem degradations of the skull could be consequent, particularly in traumatic deaths. On the other hand, the sample size is relatively small. Nonetheless, it was shown that sphenoid sinus could be used to some extent for sex identification.

Identification using paranasal sinuses three-dimensional reconstruction from CT seems to be a very popular and non-invasive perspective. However, single three- dimensional reconstructions of the sphenoid sinuses, to our knowledge, have been studied rarely. This may be due to the fact that the sphenoid sinus and ethmoid sinus are connected, so it is difficult to reconstruct the sphenoid sinus alone.

From our study, we can conclude that sphenoid sinus could be used as one of the aids for sex determination, especially if other methods used in the field of forensics are not available. Although only 62.9 % of females and 63.6 % of males were correctly classified, a high level of accuracy rate may be achieved by adding other information of sphenoid sinus. It suggests the use of MRI in forensics thus obviating the complete dependence on the usage of conventional CT and promoting the development of forensic anatomy at the soft tissue level.