Anthropometric analysis of orbital and nasal parameters for sexual dimorphism: New anatomical evidences in the field of personal identification through a retrospective observational study

Orbital and nasal parameters among modern humans show considerable variation, which affects facial shape, and these characteristics vary according to race, region, and period in evolution. The aim of the study was to ascertain whether there are sex differences in the orbital and/or nasal indexes and/or the single measurements used to calculate these in a Kosovar population. The following parameters were taken into consideration: orbital height (OH), orbital width (OW), nasal height (NH), and nasal width (NW). The ratios between orbital index/nasal index (RONI) were calculated. All measurements were obtained from a population sample comprising 408 individuals. The accuracy in sex prediction was 52.86% (CI95% = 45.05%–60.67%) for NW and 64.96% for NH (CI95% = 57.50%– 72.42%). The difference between male and female indexes was statistically significant (P < 0.05). The anthropometric study revealed that only NW and NH are configured as predictors of sexual dimorphism. It could be useful to increase the number of samples to test the discriminant function in other population groups.

a)Reply to issue no. 1: "Can you make clear why you choose female sex as a reference group?" We have added the following sentence to better explain the reason for the choice: "The female gender was chosen as the reference category because the measures of interest were lower for females and leaving this category as a reference makes the interpretation of the odds ratio findings more understandable" b)Reply to issues nos. 2 and 3: "The authors did not establish a correlation between sex difference and mean age why?" and "The authors did not establish a correlation between sex difference and mean age why?" Thanks to the reviewer for the remark. The comparison was made and reported in the 'Results' section. Below is the sentence: "251 individuals were evaluated, 110 females and 141 males, with a mean age of 37.41 (±20.56) and 33.87 (±19.58), respectively, there was no statistically significant difference in age between the sexes. (p=0.232; Student's t-test)". c)Reply to issue no. 4: "Besides forensic identification, please add some other reasons why facial measurements are necessary" In the "Discussion" section, we have added the following sentence (with reference) to better explain some other reasons why facial measurements are necessary: "Face measurements are also made for various health purposes, such as in the field of dental prostheses, cosmetic surgery, orthodontics and face masks. In particular, face analysis and proportions have been shown to be important for facial plastic surgeons to evaluate the face during the planning stages of facial reconstructive and cosmetic surgery d)Reply to issue no. 5: "Why not to compare the finding of CT scan with direct measurement of some subjects to verify whether the measurement is affected by the magnification factor or not?" Good observation, this study is in anticipation. In the conclusions section we added this sentence: "In the future, studies could be conducted to compare the measurements made on CT scans with those performed directly on the skulls and verify if there is an anthropometric variability that could make the results different" For details we invite You to check the file revised and updated in attachment. We really hope that the comments raised by the Reviewers have been satisfied in this manner. We look forward to hearing from You.  If the data are held or will be held in a public repository, include URLs, Orbital and nasal parameters among modern humans show considerable variation, which affects 26 facial shape, and these characteristics vary according to race, region, and period in evolution. The 27 aim of the study was to ascertain whether there are sex differences in the orbital and/or nasal 28 indexes (OI and NI) and/or the single measurements used to calculate these in a Kosovar 29 population. The following parameters were taken into consideration: orbital height (OH), orbital 30 width (OW), nasal height (NH), and nasal width (NW). The ratios between orbital index/nasal 31 index (RONI) were calculated. All measurements were obtained from a population sample 32 comprising 408 individuals. The sample was randomly divided into two subsamples, one  Overall, these studies have indicated some level of ethnic and racial variation in the OI and NI of 66 various population groups. In modern human groups, the appearance of the orbital and nasal 67 cavities varies considerably [5]. Craniofacial morphometric analysis is conducted using OI and NI 68 parameters, based on the ratio of the orbital/nasal height to its width multiplied by 100. These 69 ratios affect the shape of the face and vary with race, regions within the same race, and periods in 70 evolution [11][12][13][14].

71
However, to date there has been no morphometric study on OI and NI and their relationship to 72 biological sex prediction in the Kosovar population. Therefore, this study of the orbital and nasal 73 morphometry in the skulls of a Kosovar population using a CT scanner represents an important 74 contribution to the literature which will expand knowledge in various fields, such as forensic 75 medicine, and help to explore trends in evolutionary and ethnic differences, among different races 76 in particular, where forensic data is not available. Thus, the main objective of this research is to 77 ascertain whether there are sex differences in the orbital and/or nasal indexes and/or the single 78 measures used to calculate these. CT scan to investigate neurological symptoms. Exclusion criteria included growth diseases, 84 endocrine disorders or osteodystrophy, previous fractures of the of the skull and the facial massif. 85 Finally, all blurred TC images were disregarded. The CT scans were anonymously extracted and 86 limited data were extracted that were strictly necessary and relevant for the conduct of the study.

87
The extracted data are all available in the manuscript. A sample of 408 individuals was randomly 88 divided into two subsamples, one constituting approximately 2/3 (n=251) of the initial sample, and 89 the other containing approximately 1/3 of the initial sample (n=157). This division was important 90 for the steps of training (2/3 of sample) and testing (validation) (1/3 of sample) variables of interest 91 for the prediction of sex. Therefore, 251 individuals were evaluated, 110 females and 141 males, 92 with a mean age of 37.41 (±20.56) and 33.87 (±19.58), respectively. The sample of the present 93 study was calculated based on the effects observed in the study by Kotian et al. [15]. From this 94 study, the mean effect size of difference between groups of 0.20 (Cohen's d) was observed.

95
Adopting a 5% type I error (95% study confidence level), a 20% type II error (80% study power), 96 the effect size observed in the mentioned study and, considering the comparison tests of the groups 97 of the bilateral type (two-sided), it was calculated that at least 394 sample units were necessary for 98 the accomplishment of the study. As they are available, we conducted the study with 408 sample 99 units. The calculations were performed using the G*Power statistical software.

100
The study was carried out on frontal views of head CT scans performed on patients who had     Table 1 presents the distributions and comparisons of measurements for males and females.
149 Table 2 shows the results of the analysis for each predictor variable. It was not possible to fit a 150 multiple model, with more than one variable being a predictor together with another, among the 151 variables evaluated. For this reason, the univariate analysis is shown in Table 2, where it is 152 observed that, in isolation, the variables NW and NH were configured as predictors of sex. For this 153 reason, Table 2 describes only the univariate analysis.

154
In the data validation step (n=157), the accuracy of sex prediction was 52.86% for NW and 64.96% 155 for NH, considering the test cutoff points indicated (Table 3).

190
Significance level = 5%.     Finally, all blurred TC images were disregarded. The CT scans were anonymously extracted and 86 limited data were extracted that were strictly necessary and relevant for the conduct of the study.

87
The extracted data are all available in the manuscript. A sample of 408 individuals was randomly 88 divided into two subsamples, one constituting approximately 2/3 (n=251) of the initial sample, and 89 the other containing approximately 1/3 of the initial sample (n=157 units. The calculations were performed using the G*Power statistical software.

100
The study was carried out on frontal views of head CT scans performed on patients who had     In the data validation step (n=157), the accuracy of sex prediction was 52.86% for NW and 64.96% 155 for NH, considering the test cutoff points indicated (Table 3).
156 Table 3              Orbital and nasal parameters among modern humans show considerable variation, which affects 26 facial shape, and these characteristics vary according to race, region, and period in evolution. The 27 aim of the study was to ascertain whether there are sex differences in the orbital and/or nasal Finally, all blurred TC images were disregarded. The CT scans were anonymously extracted and 86 limited data were extracted that were strictly necessary and relevant for the conduct of the study.

87
The extracted data are all available in the manuscript. A sample of 408 individuals was randomly 88 divided into two subsamples, one constituting approximately 2/3 (n=251) of the initial sample, and 89 the other containing approximately 1/3 of the initial sample (n=157 units. The calculations were performed using the G*Power statistical software.

100
The study was carried out on frontal views of head CT scans performed on patients who had         found that the predominant nasal shape was mesorrhine and attributed this to environmental 258 factors, concluding that differences between nasal shapes may be used to trace geographic origins and NH). In particular, females tend to have a lower NW than that of males, as well as a higher 275 NH. However, it should be noted that, as limitation, the study has moderate accuracy with not very Thanks for the suggestion. We have deleted the previous sentence and explained our thinking better: "It was not possible to fit a multiple model, with more than one variable being a predictor together with another, among the variables evaluated. For this reason, the univariate analysis is shown in Table 2, where it is observed that, in isolation, the variables NW and NH were configured as predictors of sex" e) Reply to issue no. 5: "This is not a correct statement over here, the increase in the number of samples depends upon the sample size calculation and in this study the authors have enough sample size. however, the basis of sample taken has not been shown so the methodology clearly needs how the sample size was obtained and then the revision of the line 263 has to be thought of" The following sentence has been deleted "to increase the number of samples". The answer to criticism No. 3 shows how the sample was calculated.
The references have been corrected ***

Reviewer #2
a) Reply to issue no. 1: "Can you make clear why you choose female sex as a reference group?" We have added the following sentence to better explain the reason for the choice: "The female gender was chosen as the reference category because the measures of interest were lower for females and leaving this category as a reference makes the interpretation of the odds ratio findings more understandable" b) Reply to issues nos. 2 and 3: "The authors did not establish a correlation between sex difference and mean age why?" and "The authors did not establish a correlation between sex difference and mean age why?" Thanks to the reviewer for the remark. The comparison was made and reported in the 'Results' section. Below is the sentence: "251 individuals were evaluated, 110 females and 141 males, with a mean age of 37.41 (±20.56) and 33.87 (±19.58), respectively, there was no statistically significant difference in age between the sexes. (p=0.232; Student's t-test)".
c) Reply to issue no. 4: "Besides forensic identification, please add some other reasons why facial measurements are necessary" In the "Discussion" section, we have added the following sentence (with reference) to better explain some other reasons why facial measurements are necessary: "Face measurements are also made for various health purposes, such as in the field of dental prostheses, cosmetic surgery,