SHAPE ANALYSIS OF "INCISURA ISCHIADICA MAJOR" IN SEXING THE HUMAN PELVIS

The problem of correctly sexing unknown skeletal material is given by the contradiction between the discrete classification according to genetically determined sex and the continuous transitions of somatic sexual characters of the phenotype. This is true for pelvic bones also, despite the fact that they are the most sexually differentiated bones of the skeleton. The combination of metric characteristics of the two fundamental subsystems of os coxae (ischiopubic and sacroiliac) is not only necessary but is also sufficient for a multivariate discriminant analysis which separates completely the two sexes in a recent population. The discriminant functions obtained by using the metric characteristics of the individual pelvic segments are less successful, namely in the sacroiliac segment. To improve this result, keeping in mind the problem of sexing in unknown and partial pelvises, we have analyzed a sample of incisurae by using, instead of metric data, shape parametrical descriptions of the boundaries through Fourier analysis. In this work 97 male and 98 female os coxae were considered; the material, of known sex, was of Czech and German origin, dated to the last century (about 1880-1980) from the collections of Institutes of Anatomy in Prague and Brno. Data acquisition and treatment were performed by means of S.A.M. (Shape Analytical Morphometry) software-system; the system is constituted by an integrated architecture of analytical procedures which allow a complete description of the shape in two-dimensional objects. The boundaries to be described, dimensionally normalized and standardized for the position, are decomposed by the analysis in series of sine/cosine coefficients (from which amplitude and phase angle are obtained) for a finite number of sinusoidal harmonics; these data are then used in performing multivariate discriminant analysis to set linear functions efficacious in distinguishing the male and female groups. Having used various combinations of variables, the degree of correct attribution reached 97% of all cases. The morphological features, more informative in determining the correct sex attribution, are then identified and discussed on the basis of the obtained results.