Image analysis of pubic bone for sex determination in a computed tomography sample

Radiology has demonstrated great utility for sex determination, but most studies are based in metrical and morphological methods in order to perform an identification profile. It is presented an easy image analysis-based method aimed to demonstrate the relationship between the bony tissue ultrastructure and sex by using several grey level histogram variables obtained from computed tomography images using sagittal sections of the pubic symphysis surface and the body of the pubis. The CT sample consisted of 169 hospital DICOM archives of known sex and age. The multivariate binary logistic regression models calculated showed reliabilities for sex determination between 87.4 and 95.8 %, with a high intra and inter-observer agreement. The suggested method is not only considered useful to perform an identification profile during Virtopsy, but even for application in further studies in order to attach a quantitative correlation for tissue ultrastructure characteristics without complex and expensive methods beyond image analysis.