Many applications in orthopaedic surgery require the creation of personalised design models that can serve as the basis for navigation in computer aided surgery systems or be used to create a personalised model to perform structural analysis during pre-operative planning or post-operative follow-up. The paper introduces a method for developing a three-dimensional (3D) patient-specific model of a femur bone from an antero-posterior radiograph. A generic femur was employed and was altered on the basis of bone boundaries visible on radiographs. Morphological errors were evaluated against 3D models obtained from computed tomography (CT) scans. When only the antero-posterior radiograph was used, the average radius estimation error was 4.8 mm, the average percentage area estimation error was 14%, and the average percentage estimation error for inertial moments was 15%. If both the medial-lateral and the anterior-posterior radiographs were used, these errors were 2.0 mm, 5% and 7%, respectively. The procedure described can be profitably employed whenever CT scans are not available, such as during a retrospective analysis, or when CT scans cannot be justified because of X-ray exposure and cost considerations.
A radiograph-based femur morphing method
CRUPI, Vincenzo;
2005-01-01
Abstract
Many applications in orthopaedic surgery require the creation of personalised design models that can serve as the basis for navigation in computer aided surgery systems or be used to create a personalised model to perform structural analysis during pre-operative planning or post-operative follow-up. The paper introduces a method for developing a three-dimensional (3D) patient-specific model of a femur bone from an antero-posterior radiograph. A generic femur was employed and was altered on the basis of bone boundaries visible on radiographs. Morphological errors were evaluated against 3D models obtained from computed tomography (CT) scans. When only the antero-posterior radiograph was used, the average radius estimation error was 4.8 mm, the average percentage area estimation error was 14%, and the average percentage estimation error for inertial moments was 15%. If both the medial-lateral and the anterior-posterior radiographs were used, these errors were 2.0 mm, 5% and 7%, respectively. The procedure described can be profitably employed whenever CT scans are not available, such as during a retrospective analysis, or when CT scans cannot be justified because of X-ray exposure and cost considerations.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.