Development of the intact lumbosacral model
In this study, a healthy adult female volunteer without any history of spinal diseases was selected and the data of her CT scans (AQUIRRON 64, Toshiba, Japan, 250 mAs, 120 kV voltage, slice thickness of 0.625 mm) was obtained from the department of radiology of our hospital. The computed tomography scan images were stored in Digital Imaging and Communications in the Medicine (DICOM) format.
Anatomical 3D models of the lower lumbar vertebrae, sacrum, and coccyx were generated using Mimics research 19.0 (Materialize, Leuven, Belgium). Subsequently, the rough spinal model was imported into Geomagic Studio 2013 (3D Systems Corporation, South Carolina, USA) for further operation, including delete the spikes and the features, making triangles more uniform in size, and generate the surface model. The smoothed model was processed using SolidWorks 2017CAD (SolidWorks Corporation, Concord, MA, USA). Cortical bone, cancellous bone, nucleus pulposus, annulus fibrosus, facet cartilage, and vertebral endplates parts were constructed subsequently. The nucleus pulposus, simulated as a fluid-like and incompressible material, occupied 44% of the disc volume[9]. The thickness of the cortical bone was approximately 0.5 mm[9], and the cartilaginous endplates were modeled to be approximately 1 mm thick[10, 11]. The initial gap between the articulating surfaces was based on computed tomography images and was approximately 0.3-0.6 mm. The above parts were assembled into an intact lumbosacral model.
Three-dimensional scanning models of pedicle screw
The 3D scanner (Solutionix Rexcan CS+ 3D scanner, SolutioniX, Korea) was applied to scan and build the model of the pedicle screw. The instrument used image registration and 3D matching technology to create a point cloud of the geometric surface by a surface scanning of the target object. Then, the geometric model is generated automatically by the software. Steps to reconstructed the model were as follows: first, the surface of the fenestrated pedicle screw was sprayed with the developer evenly; after that, Ezscan 2017 software was applied to scan the fixed screws automatically; after finishing the scanning, the redundant parts of the 3D models were deleted using the lasso tool, and the file was saved in STL format.
The 3D models generated by the scanner were imported into Geomagic Studio 2013 and SolidWorks 2017CAD for further processing. Finally, the models with realistic geometry were used for the assembly of surgical models. The length and outer diameter of the pedicle screws (DePuy Synthes, California, USA) were 50 and 6.0 mm, respectively (Figure 1).
The model of bone cement
By using a random number table of CAPSI patients, a patient who was undergone fenestrated pedicle screw with cement-augmented was randomly selected from the table. The cement model was constructed by using the postoperative lumbar CT data through the above software. The volume of bone cement was approximately 1.73cm3 and distributed in a lump pattern. Then, the cement model with 1.73ml PMMA was scaled to 1 ml and 2.5 ml.
Construction of instrument models with different volumes of PMMA
The models of cage and rod were constructed in the SolidWorks 2017CAD according to the physical cage and rod. The outer diameter of the rod was 5.5mm. The length and height of the cage were 24 and 12 mm. Subsequently, unilateral transforaminal lumbar interbody fusion (TLIF) was assumed in right to remove the facet joint, facet cartilage, part of the annulus fibrosus, cartilaginous endplate, and nucleus. The screws, cement, rod, and cage were integrated with the lumbosacral model to construct six surgical models. The interbody cage is placed in the center of the intervertebral space. To control variables and maintain consistency, the consistent location of cages, screws, rods, and bone cement were used in the different surgical models (Fig. 2 and Fig. 3).
Loading and boundary conditions
All the FE model was imported into ANSYS Workbench 17.0 (ANSYS, Ltd., Canonsburg, Pennsylvania, USA ) for biomechanical testing. The material properties of cortical bone (osteoporosis), cancellous bone (osteoporosis), endplates, nucleus pulposus, annulus fibrosus, facet cartilage, cages, bone cement, and posterior spinal instrumentation was set according to previous studies (Table 1) [10, 12, 13]. The ligaments of the spine were simulated using tension-only and nonlinear spring elements [14]. The contact type of the facet joint was defined as “frictional”, and the friction coefficient was set at 0.1. The remaining bodies were defined as the “bonded” mode [11]. To reach a more accurate calculation, the tetrahedron mesh was used and the character of mesh was set up according to previous reports: the dimension of the joint cartilage mesh was 0.5 mm, while that of the other bodies was 2.0 mm. Finally, the loading and boundary conditions of the six surgical models were set up [10, 13]: The sacroiliac joint was bilaterally fixed with all degrees of moment restricted throughout the whole analysis. a vertical compressive force of 150 N was used on the upper surface of L3, and a 10 Nm moment was applied along the radial direction in flexion, extension, left lateral bending, right lateral bending, left rotation, and right rotation. The ROM, the disc stress at L3-4, and inferior articular process stress at L3 were recorded to make a biomechanical comparison of different volumes of PMMA in adjacent segments.