Analyse et modélisation statistique linéaire de la structure osseuse crânienne à partir de données morphométriques et anthropométriques pour la prédiction de paramètres mécaniques

The aim of this study is to establish and quantify the relationship between quasi-static stiffness and anthropometric and morphometric parameters of cranial osseous bones. The purpose is to predict the fracture parameters of cranial bone in crashworthiness situation. Within an experimental protocol, 289 specimens are taken from 17 unembalmed human skulls; the subjects are aged between 53 and 82. The specimen are tested under quasi-static three-points bending tests, at 10 mm/min. Morphometric data are got by analysing specimens’ photography. Values are studied by statistical analysis based on linear methods. Data are resumed by Component Principal Analysis. Groups of specimen, in function of the area of extraction of the specimen, are detected. The relationship between the stiffness and the morphometric parameters of the specimen is studied by using multiple linear regression; the input data are the weight, the density and the curvature indices of each specimen. The regression equation for the frontal and parietal areas is confirmed by residuals study; the relative error between predicted values and observed values is 3 ± 25%. There is no linear regression for the specimen taken from the temporal area of the skull.