Skip to Main content Skip to Navigation

Modélisation géométrique personalisée du membre inférieur à partir de radiographies bi-planes

Abstract : For better understanding and efficient diagnosis of musculoskeletal and osteoarticular pathologies and a more efficient diagnosis, 3D modeling of the skeleton is essential. Compared to imaging modalities such as CT-scan or MRI, the EOS system allows to acquire the 3D skeletal geometry from low dose bi-planar x-rays in standing position. To obtain these reconstructions, various methods have already been proposed and implemented in clinical routine. Nevertheless, these processes rely on a qualified operator. This work aims to automate the process, thus lowering the inter operator variability and accelerating the reconstruction with a comparable precision. Therefore, a new methodology is proposed, which relies on an initial solution based on an intuitive digitization followed by an entirely automatic optimization. This last step relies on a statistical deformation (gaussian process regression) combined with an adapted minimal path algorithm allowing automatic detection of the image contours. This fast and robust approach yields a precise 3D reconstruction in less than two minutes and has been validated in terms of shape and clinical parameters for femur, tibia, patella and pelvis. Compared to previous works, we obtained lesser or identical errors on clinical parameters within a tolerance of 1°. In addition, the approach allows better reproducibility even though the operator is a beginner. The proposed tools open the way for a more efficient 3D reconstruction of the lower limbs leading to a more accurate diagnosis.
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Wednesday, January 6, 2021 - 5:24:28 PM
Last modification on : Friday, August 5, 2022 - 2:54:01 PM
Long-term archiving on: : Wednesday, April 7, 2021 - 9:37:35 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03100690, version 1


François Girinon. Modélisation géométrique personalisée du membre inférieur à partir de radiographies bi-planes. Médecine humaine et pathologie. Ecole nationale supérieure d'arts et métiers - ENSAM, 2018. Français. ⟨NNT : 2018ENAM0068⟩. ⟨tel-03100690⟩



Record views


Files downloads