Method to Geometrically Personalize a Detailed Finite-Element Model of the Spine

Abstract

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. TBME-01357-2012.R2Abstract-To date, developing geometrically personalized and detailed solid finite element models of the spine remains a challenge, notably due to multiple articulations and complex geometries. To answer this problem, a methodology based on a free form deformation technique (kriging) was developed to deform a detailed reference finite element mesh of the spine (including discs and ligaments) to the patient-specific geometry of 10 and 82-year old asymptomatic spines. Different kriging configurations were tested: with or without smoothing, and control points on or surrounding the entire mesh. Based on the results, it is recommended to use surrounding control points and smoothing. The mean node to surface distance between the deformed and target geometries was 0.3 mm ± ± ± ± 1.1. Most elements met the mesh quality criteria (95%) after deformation, without interference at the articular facets. The method's novelty lies in the deformation of the entire spine at once, as opposed to deforming each vertebra separately, with surrounding control points and smoothing. This enables the transformation of reference vertebrae and soft tissues to obtain complete and personalized FEMs of the spine with minimal post-processing to optimize the mesh.

Quantification of Pediatric and Adult Cervical Vertebra—Anatomical Characteristics by Age and Gender for Automotive Application

“…Research on spine modeling is also increasing (Czyż et al 2012;Fice and Cronin 2012;Lalonde et al 2013;Tchako and Sadegh 2009;Zhang et al 2013). Physical and numerical models of occupants have been developed to investigate injury mechanisms.…”

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

“…Each hexagonal element was evaluated in Hypermesh (Altair Engineering, Troy, MI) for warpage, aspect ratio, skewnesss, jacobian, minimum angle, and maximum angle. The threshold criteria for the mesh quality metrics (Table 1) were determined based on experience and a comparison to other spine FE models in the literature [2], [5], [32], [43]. For robustness the automated methods were used on 90 bones, 18 full lumbar specimens, to determine how they function on a variety of bone sizes and shapes.…”

“…Several subject-specific models have been created for the lumbar spine [2][3][4][5], but the complex shape of lumbar vertebrae makes them tedious to mesh manually in large numbers. Dai et al [5] reported a semi-automated method for lumbar FE mesh creation and Lalonde et al [2] described a method based on Kriging, but both approaches require substantial manual intervention and are not easily scaled to numerous specimens.…”

“…Dai et al [5] reported a semi-automated method for lumbar FE mesh creation and Lalonde et al [2] described a method based on Kriging, but both approaches require substantial manual intervention and are not easily scaled to numerous specimens. Additionally, we are not aware of any published method for generation of lumbar spine FE models that strictly enforces feature correspondence across specimens (e.g., same node # in same anatomical position for every specimen).…”

Minimally Supervised Segmentation and Meshing of 3D Intervertebral Discs of the Lumbar Spine for Discectomy Simulation

“…Klinder et al [8] and Kelm et al [9] proposed automatic, learning-based 3D detection and segmentation frameworks of the spine, and suggested that existing disc pathologies can be reliably segmented without specifying segmentation accuracy of the pathological structure. Lalonde et al [10] proposed kriging-based deformation of a tetrahedral template mesh of the spine, which is based on high-resolution meshes that are essentially inapplicable to interactive surgery simulation as a result of a high element count. Neubert et al [11] results indicate potential of using statistical shape-aware models for segmentation of disc pathology without explicitly addressing herniated disc segmentation accuracy.…”