6th Dutch Bio-Medical Engineering Conference
26 & 27 January 2017, Egmond aan Zee, The Netherlands
10:30   Implants
Chair: Esther Tanck
15 mins
Thom Bitter, Dennis Janssen, Tim Marriott, Elaine Lovelady, Imran Khan, Nico Verdonschot
Abstract: Introduction: In total hip arthroplasty, a potential cause of failure is fretting corrosion at the head-stem modular junction, which can be simulated using Finite Element (FE) analysis. In FE analyses, different parameters can be varied to study micromotions and contact pressures at the taper interface. However, the simulation of micromotions and contact pressures in non-adaptive FE models is insufficient, as over time these can change due to interfacial change. In this study we developed an FE approach in which material removal during the wear process was simulated by adapting the taper geometry. The removal of material was validated against experiments simulating the clinical fretting wear process. Methods: An accelerated fretting screening test was developed that consistently reproduced fretting wear features observed in retrievals. Biomet Type-1 (4˚) tapers and +9 mm offset adaptors were assembled with a 4 kN force (N=3). A custom head fixture was used to create an increased offset and torque. The stems were potted in accordance with ISO 7206-6:2013. The set-up was submerged in a 37°C PBS solution with a pH adjusted to 3 using HCL and NaCl concentration of 90gl-1. The components were cyclically loaded between 0.4 - 4 kN for 10 million cycles. After completion, the volumetric and linear wear was measured using a Talyrond-585 roundness measurement machine. The FE model was created to match the experimental set up. Taper geometry and experimental material data were obtained from the manufacturer (Zimmer Biomet). The coefficient of friction of the studied combination of components was based on previous experiments (Bitter, 2016). After each change in load the geometry was updated by moving nodes inwards perpendicular to the taper surface. Archard’s Law was used to calculate the wear with the following equation: H=k*p*S, where H is the linear wear depth in mm, k is a wear factor (mm³/Nmm), p is the contact pressure (MPa) and S is the sliding distance (mm). The 10 million experimental cycles were simulated using a range of 5 to 200 computational cycles. For this purpose, the wear factor (k) was scaled for each simulation to match the volumetric wear found in the experiments. Results: The accelerated fretting experiments resulted in an average volumetric wear of 0.79 mm³ after 10 million cycles. Optimal results were found using 100 simulated cycles, and a wear factor of 1.25*10-6 (mm3/N*mm), balancing accurate results with computational time. The maximum wear depth found in the experiments was found to be 15 μm whereas the simulations predicted a maximum linear wear of 9.5 μm. Conclusion: In this study we have shown that we can accurately model wear at the taper junction. The model was validated with experiments using the measured volumetric and linear wear. With this model we will look at the effect of several patient, implant, and surgical parameters on the volumetric wear.
15 mins
Karlijn Groenen, Dennis Janssen, Yvette van der Linden, Allard Hosman, Jacky de Rooy, Jan Kooloos, Jasper Homminga, Nico Verdonschot, Esther Tanck
Abstract: Finite element models could aid in predicting the fracture risk of metastatically affected vertebrae of patients with cancer. These models should be validated against experiments. In the past, vertebral strength has mostly been established via experiments using single vertebrae [1,2]. However, more physiological loading conditions can be obtained by using three consecutive vertebrae. Therefore, the aim of our study was to develop an experimental setup and protocol able to determine vertebral bone strength in 3-segment spinal units. We obtained human cadaver specimens consisting of three consecutive complete vertebrae, two intervertebral discs, and all spinal ligaments. We created simulated metastatic lesions in half of the specimens and subsequently embedded the outer vertebrae of each specimen in PMMA. Specimens were then preconditioned and loaded in an MTS machine under axial compression at a rate of 2 mm/min until failure, while recording force and displacement. As soon as failure had occurred, the loading was paused and we fully embedded the specimens in PMMA while in this loaded state. Pre- and post-experiment CT scans were evaluated by an experienced radiologist to determine the occurrence of a fracture and/or collapse. Obtaining a clear fracture without fully destroying the specimens is important. Hence, it is important to define a clear and reliable failure criterion. In previous studies a clear drop in force (> 10-15% of peak force) has been used as failure criterion in 3-segment spinal units [3,4]. We applied this criterion to our first set of experiments (n = 4). When evaluating the post-experiment CT scans, it appeared that only two specimens showed signs of fracture. Of these, only one specimen showed a drop in force of more than 10%. In addition, in one of the specimens that did show a clear drop in force, no fracture was found. These observations suggested that occurrence of failure of the middle vertebral body cannot be assumed when using the clear (>10%) drop in force failure criterion. To ensure a successful result of the experiments, we changed the failure criterion and loaded the specimens for at least 5 mm in a second set of experiments (n = 12). Radiological examination indicated a higher success rate for this criterion, with 9 out of 12 tests in which only the middle vertebrae fractured or collapsed. Therefore, loading the specimens with at least 5 mm increased the probability of inducing fractures in the middle vertebrae. In conclusion, we developed an experimental set-up and protocol able to determine vertebral bone strength in 3-segment spinal units. In addition, this study illustrates the importance of an adequate failure criterion for successful modeling of vertebral fractures in an experimental set-up. Loading the specimens with at least 5 mm displacement, followed by a CT check, improves upon currently used failure criteria. The experiments performed will serve as input to experimentally validate finite element models we are currently developing. REFERENCES [1] J.M. Buckley et al., Spine J, 2009;9:478-485 [2] H. Ahn et al., J Spinal Disord Techn 2006;19:178-182. [3] Bürklein et al., J Biomech, 2001;35:579-587. [4] Baum et al., J Bone Miner Metab, 2014;32:56-64.
15 mins
Ashley Heuijerjans, Keita Ito, Wouter Wilson, René van Donkelaar
Abstract: Osteochondral implants are used to treat focal cartilage defects, and aim to restore joint mobility and limit further cartilage damage. Placement of the implant under an angle may have adverse effects on the treatment’s success. Implants of different mechanical properties are available and are also being developed. This can have a substantial effect on biomechanical responses of surrounding cartilage. The severity of the effects of implantation angle, implant stiffness or both combined is unclear. We hypothesize that for perfect placement, the implant material has little effect, however, a softer implant is preferred if the angle is not perfect. Finite Element simulations were performed with a component-based cartilage material model1, including viscoelastic collagen fibers in arcade architecture, and matrix porosity and swelling behavior. A non-local damage model2 simulated damage to collagen fibers (starts at 8% fiber strain) and matrix (starts at 10% deviatoric strain), softening the components. Cartilage elements were 8-node 3D with pore pressure and reduced integration. Bone and implant (8-node 3D el., red. int.) were modeled as linear elastic materials (v=0.3; Ebone = 1GPa; Eimplant = 1/10/100MPa). A 3D joint contact geometry (1 depth el.) was used, an implant was placed level at three angles (0°/5°/10°) w.r.t. the articular surface. The model was 46mm wide, cartilage 4mm thick, bone 18mm thick. The implant was 6mm wide, 10mm high. Contact was frictionless and no movement was allowed between implant and bone. Fluid in/outflow was allowed at outer edges. A 0.5 MPa compressive load was applied for 600s. A soft implant allows lateral cartilage bulging into the defect, causing internal cartilage strains. A stiffer implant does not allow such deformation and keeps internal strains minimal. Cartilage strains near softer implants are unlikely to reach critical values. However, misalignment of implantation by an angle of only 10° of a stiff implant causes high strains to opposing tissue, likely causing damage. Misaligned soft implants are predicted to cause only minor matrix damage. In conclusion, softer implants are, in accordance with the hypothesis, more forgiving to implantation angles than stiffer implants. Only when implantation is perfect, the effect on the tissue is predicted to be relatively independent on material properties. Thus, the high variability in clinical success of metal implants could be caused by misalignments of the implant. Implantation tools may aid ensuring perfect implantation angles. REFERENCES: [1] Wilson W, Huyghe JM, van Donkelaar CC, 2006. OA & Cart 14(6):554-60. [2] Hosseini SM, Wilson W, Ito K, van Donkelaar CC, 2014. OA & Cart. 22(1):95-103. ACKNOWLEDGEMENTS: This work was performed under the framework of Chemelot InSciTe.
15 mins
Thomas Anijs, Nico Verdonschot, Dennis Janssen
Abstract: Aseptic loosening is a common failure after total joint replacement procedures. Current pre-clinical testing methods of implant devices to prevent loosening are mainly focused on initial post-operative stability, but no reliable experimental methods are available to assess mid- and long-term fixation. Computation methods can be used to investigate initial stability as well as medium-/long-term bone response, under a controlled variety of conditions by changing e.g. physiological loading, bone quality and implant design and placement. In this study, hip implant fixation is assessed by considering the aspect of bone remodeling; the change in bone structure due to a difference in stresses in the bone as a result of a total hip replacement (THR). Finite-element (FE) models of a femoral bone and multiple Corail hip implant options were created; the bone model and its material properties were based on a cadaveric CT-scan. Stem placement and femoral head resection were performed following surgical instructions. Various implant sizes were tested by uniformly scaling the femur model to the implant size. Forces acting on the implanted femurs during activities of daily life (ADLs) were defined following reported muscle loading profiles of THR patients [1]. Bone remodeling outcome was subsequently computed using in-house algorithms, incorporating strain-adaptive remodeling theory [2] to determine change in bone mineral density (BMD) over post-operative time. Numerical remodeling results were based on virtually reconstructed lateral DEXA scans, to consider changes over different Gruen zones and enabling comparison with clinically measured scans. Combinations of different implant options were tested; implant size, collaring and offset option. Furthermore, effect of physiological load scaling based on Body Mass Index (BMI) following femoral size scaling is simulated. Results made clear that implant size in combination with the accordingly scaled femur had little effect on remodeling in case forces are scaled to BMI, since bone stresses were relatively the same. Use of uncorrected loads over the scaled FE models indicated increased bone loss is related to higher pre-operative bone stresses within the same bone structure. Presence of an implant collar slightly reduced BMD decrease, by increasing local bone stresses due to contact forces of the collar on the proximal end of the femur, where most of the bone loss takes place. Use of standard offset implants in an accordingly adjusted intact femur model was also found to constrain bone loss over the use of the same femur model with a corresponding high offset implant. This indicates that the hip contact force location related to a standard offset case leads to greater relative post-operative bone stresses over the entire proximal femur, improving related implant fixation.
15 mins
Florieke Eggermont, Loes Derikx, Jeffrey Free, Yvette van der Linden, Nico Verdonschot, Esther Tanck
Abstract: For the in vivo validation of patient-specific finite element (FE) models that calculate bone strength in femurs with metastatic lesions, we conducted a multi-center patient study. These patient-specific FE models are built on the basis of Quantitative Computed Tomography scans. Since the institutes in our study use different, although calibrated, CT equipment, it is essential to assess potential inter-scanner differences and their effects on subsequent FE simulations. Recently, it was shown that these differences exist and cannot be fully corrected by using a calibration phantom[1]. Also, potential changes in CT settings by deviating from the standard protocol, may affect the outcomes. Therefore, the aim of this study was to determine the effect of different CT scanners and settings on Hounsfield Units (HU), calibrated calcium equivalent values (CaEq) and FE-calculated failure load in cadaveric femurs. For this purpose, six cadaveric femurs were scanned in an anatomical body model[2] atop a solid calibration phantom on four CT scanners of three manufacturers (Philips (2x), GE and Toshiba), using the default CT protocol of the patient study: 3 mm slices, field of view (FOV) 480 mm, standard reconstruction kernel, 120 kV, variable mA, 1 s scan time, pitch < 1, in-plane resolution 0.9375 mm. Additional scans were made with variations in slice thickness (1 mm), FOV (550 mm) and reconstruction kernel (detail). For each CT scan, mean HU were calculated for a cortical (femoral shaft) and trabecular (femoral head) region of interest, which were calibrated to CaEq using the calibration phantom. Subsequently, failure loads were calculated using non-linear FE models[3]. Using the default patient protocol, mean differences between CT scanners varied up to 7% in cortical HU, 6% in trabecular HU, 6% in cortical CaEq, 12% in trabecular CaEq, and 17% in failure load. On each CT scanner, changes in slice thickness and FOV had little effect (≤ 4%), while reconstruction kernels had a large effect of maximal 11% in cortical HU, 16% in trabecular HU, 17% in cortical CaEq, 8% in trabecular CaEq, and 9% on failure load. This study confirms that quantitative analysis of CT images from different institutes should be done with care. Different equipment or scan protocols may introduce variation in HU values that needs correction before pooling imaging data and proceeding to further modeling techniques such as FEA. REFERENCES [1] Carpenter RD, et al., Medical Engineering & Physics, 2014; 36:1225-1232. [2] Tanck E, et al., Physics in Medicine & Biology, 2010; 55:N57-62. [3] Derikx LC, et al., Journal of Bone & Joint Surgery – British, 2012; 94: 1135-1142.
15 mins
Nicole van Gestel, Floor Gabriels, Jacobus Arts, Keita Ito, Sandra Hofmann, Bert van Rietbergen
Abstract: Bioactive glass (BAG) is a promising biomaterial for regeneration of large bone defects in osteomyelitis treatment, because of its bone bonding and antibacterial properties [1-3]. In the treatment of osteomyelitis with BAG granules, typically a cortical window is created and the necrotic and infected bone is debrided, resulting in a large bone defect. As a result, the stiffness and strength of the bone can be severely reduced. The goal of this study was to experimentally measure to what extent the stiffness of the bone is reduced after creating a typical defect and if this stiffness can be (partially) restored when filling the defect with BAG granules [4]. The stiffness of nine excised sheep radii of 100 mm in length was determined by four point bending tests. Loading was applied at a speed of 0.05 mm/s until a bending moment of 3.75 Nm was reached. All tests were executed in triplicate and the stiffness was calculated from the slope of the last part of the force-displacement curve, as averaged for the last two tests. Subsequently, a cortical window was created in the center part of the radii at the side loaded in compression during the bending tests, and the underlying bone was reamed. The defect was made similar in size relative to the bone diameter as in human applications. The four-point bending tests then were repeated for the bone with defect. Finally, the defect was filled with BAG while a single parafilm layer sealed the defect and the bending tests were repeated. The stiffnesses of the conditions were pairwise compared and the results showed a significant decrease in stiffness of 13.3 % (SD 6.13 %) when a cortical window was created (p-value < 0.001). This decrease was significantly reversed by reconstruction with BAG granules after which the stiffness was 96.6 % (SD 5.76 %) of the intact stiffness (p-value = 0.082). These results thus show that the defect can severely reduce the bone stiffness and that the BAG granules are able to restore most of this lost stiffness. This indicates that BAG can be load-bearing in clinically-confined situation, at least directly post-operative. References [1] Ö. H. Andersson et al. “In vivo behaviour of glasses in the SiO2-Na2O-CaO-P2O5-Al2O3-B2O3 system,” J. Mater. Sci. Mater. Med., vol. 1, no. 4, pp. 219–227, Nov. 1990. [2] E. Munukka et al. “Bactericidal effects of bioactive glasses on clinically important aerobic bacteria.,” J. Mater. Sci. Mater. Med., vol. 19, no. 1, pp. 27–32, Jan. 2008. [3] N. A. P. van Gestel et al. “Clinical Applications of S53P4 Bioactive Glass in Bone Healing and Osteomyelitic Treatment: A Literature Review,” Biomed Res. Int., vol. 2015, pp. 1–12, 2015. [4] N. A. P. van Gestel et al. “Bioactive glass can potentially reinforce large bone defects,” in Front. Bioeng. Biotechnol. Conference Abstract: 10th WBC, 2016.