Normal finite element model
We selected a healthy male volunteer to generate a normal finite element model. The image data in DICOM format of five vertebrae and four discs between T10 and L2 were obtained with a 64-slice spiral computed tomography scanner (Siemens, Erlangen, Germany) at 1-mm interlayer spacing. We imported the images into Mimics v20.0 (Materialise, Leuven, Belgium) to create 3D vertebral surface models of T10 to L2 in STL format. The posterior structure and intervertebral discs (IVDs) were constructed using 3-matic v12.0 (Materialise) [17-19]. The models were imported into Geomagic Studio v12.0 (Geomagic, Research Triangle Park, NC, USA) and processed by smoothing and surface and grid construction. The bone and ligament structures were meshed using Hypermesh 2017 (Altair Engineering, Troy, MI, USA). Abaqus 2019 (Simulia, Johnston, RI, USA) was used for material property definition, model assembly, loading, and finite element analysis. The intact T10–L2 finite element model is shown in Figure 1. After mesh convergence, the mesh sizes of the vertebral body and IVD were 1.5 and 1 mm, respectively. The cortical bone, facet joint, and cartilage endplate were simulated with shell elements with thicknesses of 1, 0.2, and 0.5 mm, respectively [17,18]. IVDs were divided into nucleus pulposus, annulus fibrosus, and endplates. The nucleus pulposus accounts for 30–40% of intervertebral volume. The annulus fibrosus is composed of the annulus fibrosus matrix and fibers that are divided into three–five layers at an angle of 30° [17,19]. Seven ligaments (anterior longitudinal, posterior longitudinal, interspinous, supraspinous, intertransverse, and capsular ligaments and ligamentum flavum) were created for each segment [17]. Ligaments and fibers were simulated using T3D2 elements.
Implants and fixation models
SolidWorks (Dassault Systemes, Paris, France) was used to draw pedicle screws (6.5×45 mm, 6.0×40 mm), rods (5.5 mm), and a 3D-printed prosthesis with an artificial pedicle structure (AK Medical, Beijing, China) (35×20×15 mm) according to the size of the 3D finite element model, which was validated using Geomagic Studio v12.0. The material properties used in the finite element model (Table 1) were based on previous reports [20,21].
Table 1. Material properties for the thoracolumbar spine finite element model
Structure
|
Young’s modulus (MPa)
|
Poisson ratio
|
Cross-sectional area (mm2)
|
Vertebrae
|
|
|
|
Cancellous bone
|
100
|
0.2
|
|
Cortical bone
|
12,000
|
0.3
|
|
Posterior elements
|
3500
|
0.25
|
|
Disc
|
|
|
|
Annulus
|
4.2
|
0.45
|
|
Nucleus
|
0.2
|
0.49
|
|
Facet
|
11
|
0.2
|
|
Ligaments
|
|
|
|
Anterior longitudinal ligament
|
7
|
|
63.7
|
Posterior longitudinal ligament
|
7
|
|
20
|
Ligamentum flavum
|
3
|
|
40
|
Intratransverse ligament
|
7
|
|
1.8
|
Capsular ligament
|
4
|
|
30
|
Interspinous ligament
|
6
|
|
40
|
Supraspinous ligament
|
6.6
|
|
30
|
Pedicle screw and rod fixation
|
110,000
|
0.3
|
|
3D-printed prosthesis
|
675
|
0.3
|
|
A T12 TES model was created using 3-matic v12.0 with the whole vertebra and adjacent IVD (T11/12, T12/L1 IVD) removed. Two surgical models were constructed (Fig. 1), each using a different combination of screws. In Model A, the 3D-printed prosthesis was fixed with long-segment posterior fixation(with two-level pedicle screwes fixation above and below the VBR; T10/11 and L1/2) and in Model B, artificial pedicle screws connected the 3D-printed prosthesis in Model A (T10/T11, L1/2, and the 3D-printed prosthesis). The pedicle screws inserted into the vertebral body were 6.0×40 mm, and the pedicle screws inserted into the 3D-printed prosthesis were 6.5×45 mm.
Boundary and loading conditions
Abaqus 2019 was used to set boundary and load conditions and simulate spinal movement. We assumed that the L2 vertebral body was fixed, and its substructure was set as a boundary with no displacement or rotation in any direction. Spinal motion in sagittal, coronal, and cross-sections was defined as flexion, extension, lateral bending, and rotation. We applied an axial load of 200 N and torque load of 7.5 N·m to the upper surface coupling point of T10 to simulate the flexion, extension, lateral bending, and rotation of the spine [25,26].
Assessment indices
Three indices were used to assess the mechanical properties of the structure: stiffness of the construct (T10–L2), von Mises stress of the internal fixed system, and von Mises stress on the endplate adjacent to the 3D-printed prosthesis (L1 superior endplate). We used these indices to evaluate the biomechanical effects of the artificial pedicle in the constructed models. Since only one subject was modeled, no statistical analysis was performed in this study.