Abstract
To validate the use of panoramic radiographs and morphometric parameters in forensic odontology for accurate and efficient gender determination in the specific socio-cultural context of the Pakistani population.A retrospective study was conducted using orthopantomograms from 130 individuals aged between 16 and 30 years, sourced from a radiology department. The study included comprehensive measurements of mandibular indices such as condylar height, coronoid height, and antegonial depth. Measurements were made using Image J software. The indices were analyzed through univariate, multivariate, and best models to assess their effectiveness in gender prediction. Statistical analysis included independent samples t-test, binary logistic regression, and receiver operator characteristic (ROC) analysis to evaluate threshold values, sensitivity, specificity, and area under the curve (AUC) for each index.Independent samples t-test was used to compare the means of indices with gender. Binary logistic regression was used to estimate the likelihood of male gender, and ROC analysis was used to calculate threshold values, sensitivity, specificity, and AUC.Univariate analysis revealed that most indices, except for the gonial angle, showed significant differences between genders. The multivariate model stated the condylar height and coronoid height as a significant predictor. The best model confirmed condylar height, coronoid height, antegonial depth, and the inferior border of the mental foramen as reliable indices for male gender determination. The ROC demonstrated that the distance from the mean inferior border to the lower border of the mandible had the highest AUC of 82%, indicating strong predictive power.The study confirmed the effectiveness of specific mandibular measurements in gender determination within the Pakistani population. Condylar height, coronoid height, antegonial depth, and the inferior border of the metal foramen are consistently significant predictors across various models. Further research with a larger population sample is recommended.