Anatomical comparison and evaluation of human proximal femurs modeling via different devices and FEM analysis

Abstract

Breuckmann optical scanning, Metris laser scanning and CT are general devices used for modeling hard or soft tissues in the biomedical field. Whereas the CT device is able to model internal and external structures of hard tissues, Breuckmann and Metris devices can only model the exterior portions of tissues. In this study, a human proximal femur was modeled using these devices, and the matching accuracy thereof is presented. The human proximal femur was modeled by scanning with Breuckmann optical scanning, Metris laser scanning and CT devices. The 3D/3D registration method was performed in two ways: coordinate to coordinate based, and 2D contour based matching. Matching accuracies of the three models were developed with statistical deviation and local deviation. To determine the significance value between the deviations obtained, one way ANOVA, and for intragroup comparisons Tukey and Thamhane tests were used. After statistical analysis, stresses on the models were evaluated using ANSYS software taking boundary conditions on human standing position into consideration. In this study, the value of the 2D contour based accuracy deviation of the femur head zone between CT and Metris models was obtained as 0.4 ± 0.2 mm while it was 0.3 ± 0.1 mm between CT and Breuckmann. The highest matching deviation obtained as a result of the ANOVA test among these three models was found in the femur trochanter region (0.0142 ± 0.0164 mm), the lowest value was found in the femur head region (0.0070 ± 0.0132 mm). The stress of the CT-Breuckmann pair was found close to each other in stress analysis. The deviation values obtained by matching models created by three different methods showed statistically significant results (P < 0.05). Values obtained from the CT-Breuckmann model were lower than those obtained from CT-Metris. In order to lower deviation values, applications such as increasing the resolution of images, using stronger algorithms, meshing methods and enhancing surface form should be implemented.