An Automated Method for Landmark Identification and Finite-Element Modeling of the Lumbar Spine

Goal: The purpose of this study was to develop a method for the automated creation of finite-element models of the lumbar spine. Methods: Custom scripts were written to extract bone landmarks of lumbar vertebrae and assemble L1-L5 finite-element models. End-plate borders, ligament attachment points, and facet surfaces were identified. Landmarks were identified to maintain mesh correspondence between meshes for later use in statistical shape modeling. Results: 90 lumbar vertebrae were processed creating 18 subject-specific finite-element models. Finite-element model surfaces and ligament attachment points were reproduced within 1e-5 mm of the bone surface, including the critical contact surfaces of the facets. Element quality exceeded specifications in 97% of elements for the 18 models created. Conclusion: The current method is capable of producing subject-specific finite-element models of the lumbar spine with good accuracy, quality, and robustness. Significance: The automated methods developed represent advancement in the state of the art of subject-specific lumbar spine modeling to a scale not possible with prior manual and semiautomated methods.