ABSTRACT
Objective Logistic regression method was used to establish a multiple regression sex discriminant function to discriminate the complete skull model and the incomplete skull model without frontal bone, occipital bone and mandible of Uygur adults in Turpan, Xinjiang. Methods A total of 117 （60 male and 57 female） three-dimensional skull models were collected by CT. Sixteen cranial measurement indexes were measured and calculated by computer software. The multivariate regression sex discriminant function was established with Logistic regression method and retrospectively tested. Results Among the 16 measurement indexes, except for nose width （x7） and maximum frontal breadth （x13）, the remaining 14 indexes had statistical significance of differences between male and female （P<0.05）. For the discriminant function of complete skull established by eyebrow arch convexity （x4）, mastoid width （x6）, maximum cranial length （x12）, cranial base length （x15）, cranial circumference （x16）, the male and female discrimination accuracy was 90.0% and 94.7%, respectively. For the sex discriminant function of incomplete skull without frontal bone established by mandibular angle width （x10）, mandibular height （x11） and cranial circumference （x16）, the discrimination accuracy of male and female was 85.0% and 84.2%, respectively. For the sex discriminant function of incomplete skull without occipital bone established by the index of eyebrow arch convexity （x4）, the discrimination accuracy of male and female was 80.0% and 73.7%, respectively. For the sex discriminant function of incomplete skull without mandible established by frontal chord （x5） and occipital protrusion angle （x9）, the discrimination accuracy of male and female was 85.0% and 78.9%, respectively. Conclusion The computer software and system developed in our study can achieve sex discrimination of complete skulls and incomplete skulls without frontal bone, occipital bone or mandible.