Purpose. As human soft tissue is anisotropic, non-linear and inhomogeneous, its properties are difficult to characterize. Different methods have been described that are either based on contact or noncontact protocols. In this study, three-dimensional (3D) digital image correlation (DIC) was adopted to examine the mechanical behaviour of the human Achilles tendon. Despite its wide use in engineering research and its great potential for strain and displacement measurements in biological tissue, the reported biomedical applications are rather limited. To our knowledge, no validation of 3D DIC measurement on human tendon tissue exists. The first goal of this study was to determine the feasibility to evaluate the mechanical properties of the human Achilles tendon under uniaxial loading conditions with 3D Digital Image Correlation. The second goal was to compare the accuracy and reproducibility of the 3D DIC against two linear variable differential transformer (LVDT's). Methods. Six human Achilles tendon specimens were prepared out of fresh frozen lower limbs. Prior to preparation, all limbs underwent CT-scanning. Using Mimics software, the volume of the tendons and the cross sectional area at each level could be calculated. Subsequently, the Achilles tendons were mounted in a custom made rig for uni-axial loading. Tendons were prepared for 3D DIC measurements with a modified technique that enhanced contrast and improved the optimal resolution. Progressive static loading up to 628,3 N en subsequent unloading was performed. Two charge-coupled device camera's recorded images of each loading position for subsequent strain analysis. Two LVDT's were mounted next to the clamped tendon in order to record the displacement of the grips. Results. 3D DIC strain measurement proved to be technical feasible on human tendon tissue if an adapted preparation protocol is used. A spatial resolution of 0,1 mm was reached. Accuracy analysis shows a very low scatter, comparable to that obtained in steel applications (0,03%). When compared to the LVDT measurements, DIC showed excellent correlation (R = 0.99). Apart from the longitudinal strain component, an important transverse strain component was revealed in all specimens (fig 1). Also a significant amount of slip was observed at the clamps. Through the non-contact nature of the measurement, this could be quantified and the analysis became independent of any slip (fig 2). The strain distribution was of a strongly inhomogeneous nature, both within the same specimen (fig 1) and amongst different specimens. Conclusion. 3D DIC is a very promising technique for strain measurement of human collagenous tissue. Accuracy analyses indicate a very low scatter, comparable to that obtained in traditional steel applications. The major advantages of the DIC technique over the LVDTs is the 3D, non-contact, full-field nature of the analysis and the possibility to analyse multidirectional strain, without disturbing slip effects in the grips

Introduction. In this study, three-dimensional (3D) digital image correlation (DIC) was adopted to investigate the strain in the superficial medial collateral ligament (sMCL) of the human knee. To our knowledge, no reports or validation of 3D DIC measurement on human collagenous tissue exists. The first part of this research project focused on the validation of 3D DIC (1) as a highly accurate tool for non-contact full field strain analysis of human collagenous tissue. In the second part, 3D DIC was used to measure the strain patterns in the superficial medial collateral ligament (sMCL) of the native knee (2). In a third part, the strain pattern in the sMCL after total knee arthroplasty (TKA) in an ‘optimal’ (3) and with a proximalised joint line (4) was analysed. Methods. (1) Six fresh frozen human Achilles tendon specimens were mounted in a custom made rig for uni-axial loading. The accuracy and reproducibility of 3D DIC was compared to two linear variable differential transformers (LVDT's). (2) The strain pattern of the sMCL during the range of motion (ROM) was measured using 3D DIC in six fresh frozen cadaveric knees. The knees were mounted in a custom made rig, applying balanced tension to all muscle groups around the knee. The experiment was repeated after computer navigated implantation of a single radius posterior stabilised (PS) TKA in ‘optimal’ (3) and with a 4 mm proximalised joint line (4). Results. (1) Accuracy analysis revealed that the scatter was very low for all specimens (0,03%) and a spatial resolution of 0,1 mm for strain measurement was reached. When compared to the LVDT, DIC showed excellent correlation (R = 0.99). (2) Overall, the sMCL behaved isometrically between 0° and 90° of flexion showing less 1% slackening in all specimens. Further slackening was seen in deeper flexion. Significant regional inhomogeneity was observed (fig 1). The highest strains (up to 5% lengthening) were seen in the proximal part. The middle and distal part were near isometric between 0° and 90° of flexion. (3) A significant alteration of the strain pattern was seen after TKA with an increased strain in all parts of the sMCL from 90° to deeper flexion (fig 2). (4) This effect became significantly more pronounced with joint line proximalisation. Discussion. Strain in the native sMCL proved to be inhomogeneously distributed with significant differences between proximal, middle and distal part during the ROM. The higher baseline strain in the proximal part might be the explanation for the fact that most of the sMCL lesions are seen in that region. A single radius TKA failed to reproduce the native sMCL strain pattern from 90° to deeper flexion. This effect became even more pronounced with joint line proximalisation. These higher sMCL strains might compromise deeper flexion after TKA. Conclusion. The strain pattern of the sMCL in the native knee showed important regional differences during the ROM and significant alterations after TKA implantation and joint line proximalisation

Introduction. Periprosthetic bone remodelling after Total Knee Arthroplasty (TKA) may be attributed to local changes in the mechanical strain field of the bone as a result of the stiffness mismatch between high modulus metallic implant materials and the supporting bone. This can lead to significant loss of periprosthetic bone density, which may promote implant loosening, and complicate revision surgery. A novel polyetheretherketone (PEEK) implant with a modulus similar to bone has the potential to reduce stress shielding whilst eliminating metal ion release. Numerical modelling can estimate the remodelling stimulus but rigorous validation is required for use as a predictive tool. In this study, a finite element (FE) model investigating the local biomechanical changes with different TKA materials was verified experimentally using Digital Image Correlation (DIC). DIC is increasingly used in biomechanics for strain measurement on complex, heterogeneous anisotropic material structures. Methodology. DIC was used following a previously validated technique [1] to compare bone surface strain distribution after implantation with a novel PEEK implant, to that induced by a contemporary metallic implant. Two distal Sawbone® femora models were implanted with a cemented cobalt-chromium (CoCr) and PEEK-OPTIMA® femoral component of the same size and geometry. A third, unimplanted, intact model was used as a reference. All models were subjected to standing loads on the corresponding UHMWPE tibial component, and resultant strain data was acquired in six repeated tests. An FE model of each case, using a CT-derived bone model, was solved using ANSYS software. Results and Discussion. The sensitivity of DIC strain measurements was <+130με and experimental error was +230με, or 8.5% of the peak magnitude in the region of interest. High bone strain adjacent to the CoCr implant and low bone strain in the central metaphyseal region compared to the intact case (Fig.1) indicated that stress shielding may lead to resorption, a theory corroborated by bone density scans of implanted metallic TKRs [2]. Quantitatively, wider scatter and greater deviation was observed between the intact-vs-CoCr datasets (R. 2. : 0.425, slope = 0.508). A closer agreement was shown between the intact-vs-PEEK datasets (R. 2. : 0.771, slope = 1.270) (Fig.2). These strain distributions corroborated the predictions of the FE analysis (Fig.1). High bone strain in regions close to the CoCr implant can be attributed to the high stiffness mismatch between implant and bone, where the bone is constrained to the implant with cement. High strain gradients near the stiff CoCr could potentially compromise implant fixation, leading to loosening. The compressive strains in the PEEK implanted model were similar to those in the intact case, suggesting that bone would be maintained in these regions, and high strain gradients were not observed. Conclusion. Digital image correlation and FE analysis have been successfully employed for evaluation of a novel PEEK-OPTIMA® TKA implant in comparison to a metallic implant. The polymeric implant produced a strain distribution closer to that of the intact bone, and therefore would be expected to have less of a stress shielding effect, improving long term bone preservation

INTRODUCTION. In theory, Finite Element Analysis (FEA) is an attractive method for elucidating the mechanics of modular implant junctions, including variations in materials, designs, and modes of loading. However, the credence of any computational model can only be established through validation using experimental data. In this study we examine the validity of such a simulation validated by comparing values of interface motion predicted using FEA with values measured during experimental simulation of stair-climbing. MATERIALS and METHODS. Two finite element models (FEM) of a modular implant assembly were created for use in this study, consisting of a 36mm CoCr femoral head attached to a TiAlV rod with a 14/12 trunnion. Two head materials were modelled: CoCr alloy (118,706 10-noded tetrahedral elements), and alumina ceramic (124,710 10-noded tetrahedral elements). The quasi-static coefficients of friction (µ. s. ) of the CoCr-TiAlV and Ceramic-TiAlV interfaces were calculated from uniaxial assembly (2000N) and dis-assembly experiments performed in a mechanical testing machine (Bionix, MTS). Interface displacements during taper assembly and disassembly were measured using digital image correlation (DIC; Dantec Dynamics). The assembly process was also simulated using the computational model with the friction coefficient set to µ. s. and solved using the Siemens Nastran NX 11.0 Solver. The frictional conditions were then varied iteratively to find the value of µ providing the closest estimate to the experimental value of head displacement during assembly. To validate the FEA model, the relative motion between the head and the trunnion was measured during dynamic loading simulating stair-climbing. Each modular junction was assembled in a drop tower apparatus and then cyclically loaded from 230–4300N at 1 Hz for a total of 2,000 cycles. The applied load was oriented at 25° to the trunnion axis in the frontal plane and 10° in the sagittal plane. The displacement of the head relative to the trunnion during cyclic loading was measured by a three-camera digital image correlation (DIC) system. The same loading conditions were simulated using the FEA model using the optimal value of µ derived from the initial head assembly trials. RESULTS. For both head materials, the predicted values of axial displacement of the head on the trunnion closely approximated the measured values derived from DIC measurements, with differences of −0.17% to +6.5%, respectively. Larger differences were calculated for individual components of motion for the stair climbing activity. However, the predicted magnitude of interface motion was still within 10% of the observed values, ranging from −7% to −5%. CONCLUSIONS. Our simulations closely approximated physical testing using complex loading, coming within 7% of the target values. By generating a validated computational model of a modular junctions with varying head materials, we will be able to simulate additional activities of daily living to determine micromotion and areas of peak pressure and contact stresses generated. For any figures or tables, please contact authors directly

Introduction. Polyetheretherketone (PEEK) has been proposed as an implant material for femoral total knee arthroplasty (TKA) components. Potential clinical advantages of PEEK over standard cobalt chrome alloys include modulus of elasticity and subsequently reduced stress shielding potentially eliminating osteolysis, thermal conduction properties allowing for a more natural soft tissue environment, and reduced weight enabling quicker quadriceps recovery. Manufacturing advantages include reduced manufacturing and sterilization time, lower cost, and improved quality control. Currently, no PEEK TKA implants exist on the market. Therefore, evaluation of mechanical properties in a pre-clinical phase is required to minimize patient risk. The objectives of this study include evaluation of implant fixation and determination of the potential for reduced stress shielding using the PEEK femoral TKA component. Methods and Materials. Experimental and computational analysis was performed to evaluate the biomechanical response of the femoral component (Freedom Knee, Maxx Orthopedics Inc., Plymouth Meeting, PA; Figure 1). Fixation strength of CoCr and PEEK components was evaluated in pull-off tests of cemented femoral components on cellular polyurethane foam blocks (Sawbones, Vashon Island, WA). Subsequent testing investigated the cemented fixation using cadaveric distal femurs. The reconstructions were subjected to 500,000 cycles of the peak load occurring during a standardized gait cycle (ISO 14243-1). The change from CoCr to PEEK on implant fixation was studied through computational analysis of stress distributions in the cement, implant, and the cement-implant interface. Reconstructions were analyzed when subjected to standardized gait and demanding squat loads. To investigate potentially reduced stress shielding when using a PEEK component, paired cadaveric femurs were used to measure local bone strains using digital image correlation (DIC). First, standardized gait load was applied, then the left and right femurs were implanted with CoCr and PEEK components, respectively, and subjected to the same load. To verify the validity of the computational methodology, the intact and reconstructed femurs were replicated in FEA models, based on CT scans. Results. The cyclic load phase of the pull-off experiments revealed minimal migration for both CoCr and PEEK components, although after construct sectioning, debonding at the implant-cement interface was observed for the PEEK implants. During pull-off from Sawbones the ultimate failure load of the PEEK and CoCr components averaged 2552N and 3814N respectively. FEA simulations indicated that under more physiological loading, such as walking or squatting, the PEEK component had no increased risk of loss of fixation when compared to the CoCr component. Finally, the DIC experiments and FEA simulations confirmed closer resemblance of pre-operative strain distribution using the PEEK component. Discussion. The biomechanical consequences of changing implant material from CoCr to PEEK on implant fixation was studied using experimental and computational testing of cemented reconstructions. The results indicate that, although changes occur in implant fixation, the PEEK component had a fixation strength comparable to CoCr. The advantage of long term bone preservation, as the more compliant PEEK implant is able to better replicate the physiological loads occurring in the intact femur, may reduce stress shielding around the distal femur, a common clinical cause of TKA failure. For any figures or tables, please contact the authors directly

Introduction. Aseptic loosening of total knee replacements is a leading cause for revision. It is known that micromotion has an influence on the loosening of cemented implants though it is not yet well understood what the effect of repeated physiological loading has on the micromotion between implants and cement mantle. This study aims to investigate effect of physiological loading on the stability of tibial implants previously subjected to simulated intra-operative lipid/marrow infiltration. Methods. Three commercially available fixed bearing tibial implant designs were investigated in this study: ATTUNE. ®. , PFC SIGMA. ®. CoCr, ATTUNE. ®. S+. The implant designs were first prepared using a LMI implantation process. Following the method described by Maag et al tibial implants were cemented in a bone analog with 2 mL of bone marrow in the distal cavity and an additional reservoir of lipid adjacent to the posterior edge of the implant. The samples were subjected to intra- operative range of motion (ROM)/stability evaluation using an AMTI VIVO simulator, then a hyperextension activity until 15 minutes of cement cure time, and finally 3 additional ROM/stability evaluations were performed. Implant specific physiological loading was determined using telemetric tibial implant data from Orthoload and applying it to a validated FE lower limb model developed by the University of Denver. Two high demand activities were selected for the loading section of this study: step down (SD) and deep knee bend (DKB). Using the above model, 6 degree of freedom kinetics and kinematics for each activity was determined for each posterior stabilized implant design. Prior to loading, the 3-D motion between tibial implant and bone analog (micromotion) was measured using an ARAMIS Digital Image Correlation (DIC) system. Measurement was taken during the simulated DKB at 0.25Hz using an AMTI VIVO simulator while the DIC system captured images at a frame rate of 10Hz. The GOM software calculated the distance between reference point markers applied to the posterior implant and foam bone. A Matlab program calculated maximum micromotion within each DKB cycle and averaged that value across five cycles. The implant specific loading parameters were then applied to the three tibial implant designs. Using an AMTI VIVO simulator each sample was subjected to 50,000 DKB and 120,000 SD cycles at 0.8Hz in series; equating to approximately 2 years of physiological activity. Following loading, micromotion was measured using the same method as above. Results. Initial micomotion measurements during DKB activity for ATTUNE. ®. , PFC SIGMA. ®. CoCr, ATTUNE. ®. S+ were 155µm, 246µm, and 104µm, respectively, and following physiological loading were 159µm, 264µm, and 112µm, respectively. While there was statistical significance between the micromotion of implant designs (p<0.05), there was no significance between before and after loading. Conclusion. This study shows there is no significant change in micromotion after approximately 2 years of physiological loading. However, there is a significant difference in micromotion between implant designs

Introduction. Although cementless press-fit femoral total knee arthroplasty (TKA) components are routinely used in clinical practice, the effect of the interference fit on primary stability is still not well understood. Intuitively, one would expect that a thicker coating and a higher surface roughness lead to a superior fixation. However, during implant insertion, a thicker coating can introduce more damage to the underlying bone, which could adversely influence the primary fixation. Therefore, in the current study, the effect of coating thickness and roughness on primary stability was investigated by measuring the micromotions at the bone-implant interface with experimental testing. Methods. A previous experimental set-up was used to test 6 pairs of human cadaveric femurs (47–60 years, 5 females) implanted with two femoral component designs with either the standard e.motion (Total Knee System, B. Braun, Germany) interference fit of 350 µm (right femurs) or a novel, thicker interference fit of 700 µm (left femurs). The specimens were placed in a MTS machine (Figure 1) and subjected to the peak loads of normal gait (1960N) and squat (1935N), based on the Orthoload dataset for Average 75. Varus/valgus moments were incorporated by applying the loads at an offset relative to the center of the implants, leading to a physiological mediolateral load distribution. Under these loads, micromotions at the implant-bone interface were measured using Digital Image Correlation (DIC) at different regions of interest (ROIs – Figure 1). In addition, DIC was used to measure opening and closing of the implant-bone interface in the same ROIs. Results. After comparing the micromotions and opening of the two implant designs, we found no significant differences between the standard and novel coating. Loading was a significant factor for both opening (P<0.0001) and micromotions (P=0.019), where the squat produced higher micromotions than gait. Opening was seen anteriorly (MA, LA), and was higher during squat. Closing was noticed distally (MD, LD), particularly during gait (Figure 2). During gait (Figure 3), the highest micromotions were found in the posterior condyles (CM, MP), followed by the medial anterior region (MA). For squat, the largest micromotions were in the anterior flange (ANT), followed by the distal regions (LD, MD). Discussion. In the current study, the primary stability of the same implant with two different coating thicknesses was evaluated. The results demonstrate that increasing the coating thickness does not automatically influence the primary stability of a femoral TKA component. This is likely due to abrasion and damage of the underlying trabecular during implant insertion, which also was observed in previous experiments. The exact relation between coating thickness or interference fit and primary implant stability still remains subject to debate. Obviously, the primary implant stability is compromised when the interference fit is too low. However, the current results suggest that there is a threshold beyond which further improvement of the fixation is not possible. The exact magnitude of this threshold is unknown, and may depend on coating characteristics and bone quality, and requires further evaluation, possibly utilizing a hybrid approach of experimental and computational techniques

INTRODUCTION. Mechanical overloading of the knee can occur during activities of daily living such as stair climbing, jogging, etc. In this finite element study we aim to investigate which parameters could detrimentally influence peri-implant bone in the tibial reconstructed knee. Bone quality and patient variables are potential factors influencing knee overloading (Zimmerman 2016). METHODS. Finite element (FE) models of post-mortem retrieved tibial specimens (n=7) from a previous study (Zimmerman 2016) were created using image segmentation (Mimics Materialise v14) of CT scan data (0.6 mm voxel resolution). Tibial tray and polyethylene inserts were recreated from CT data and measurements of the specimens (Solidworks 2015). Specimens with varying implant geometry (keel/pegged) were chosen for this study. A cohesive layer between bone and cement was included to simulate the behavior of the bone–cement interface using experimentally obtained values. The FE models predict plasticity of bone according to Keyak (2005). Models were loaded to 10 body weight (BW) and then reduced to 1 BW to mimic experimental measurements. Axial FE bone strains at 1 BW were compared with experimental Digital Image Correlation (DIC) bone strains on cut sections of the specimens. After validation of the FE models using strain data, models were rotated and translated to the coordinate system defined in Bergmann (2014). Four loading cases were chosen – walking, descending stairs, sitting down and jogging. Element strains were written to file for post-processing. The bone in all FE models was divided into regions of equal thickness (10 mm) for comparison of strains. RESULTS. Results are shown for two specimens at present. Strain-maps of the specimen cut section compare reasonably well with FE cutting-plane strains. The FE models however show some regions of high strain in certain locations which do not correspond with the experimental results (Figure 1). Plasticity predicted by the models at 10 BW is shown in Figure 2. Median bone strains for two loading cases are shown as a function of distance below the tibial tray in Figure 3. This figure shows that specimen 1 is less likely to be overloaded during jogging when compared with specimen 2. Both specimens remain below the 7300 με threshold for compressive yield. DISCUSSION. Using functioning knee replacement tibial specimens, we study which factors influence bone overloading. Validation using DIC strain measurements is challenging due to the large plasticity regions predicted by the material model used here. The present results were obtained using plasticity relationships from Keyak (2005) for the proximal femur. To further improve on our results, plasticity-bone density relationships for the proximal tibia (Keyak 1996) will be included. Proximal tibial bone has been shown to be stiffer than femoral bone (Morgan 2003). Despite these limitations, FE models provide valuable information on the risk of overloading during daily living activities. The study will be expanded to include an analysis of implant geometry, bone quality and other loading cases. For any figures or tables, please contact the authors directly

It is nowadays widely recognized that patient satisfaction following knee arthroplasty strongly depends on ligament balancing. To obtain this balancing, the occurring ligament strain is assumed to play a crucial role. To measure this strain, a method is described in this paper that allows full field 3D evaluation of the strains. The latter is preferred over traditional measurement techniques, e.g. displacement transducers or strain gauges, as human soft tissue is not expected to deform uniformly due to its highly inhomogeneous and anisotropic properties. To facilitate full field strain measurements, the 3D digital image correlation (DIC) technique was adopted. This technique was previously validated by our research group on human tissue. First, a high contrast speckle pattern was applied on the sMCL. Therefore, the specimens are first coated with a small layer of methylene blue. Following, a random white speckle pattern is applied. During knee flexion, two cameras simultaneously take pictures of the deforming region at predefined flexion angles. Using dedicated software, the captured images are eventually combined and result in 3D full field strains and displacements. Using this method, the strain distribution was studied in six cadaveric knees during flexion extension movement. Therefore, the femur was rigidly fixed in a custom test rig. The tibia was left unconstrained, allowing the six degrees of freedom in the knee. A load was applied to all major muscles in physiological directions of each muscle by attaching a series of calibrated weights (Farahmand et al., J Orthop Res., 1998;16(1)). The direction of the pulling cables was controlled using a digital inclinometer for each specimen. As a result, a statically balanced muscle loading of the knee was obtained. From these cadaveric experiments, it is observed that on average the sMCL behaves isometrically between 0° and 90° of flexion. However, high regional differences in strain distribution are observed from the full field measurements. The proximal region of the sMCL experiences relatively high strains upon flexion. These strains are positive (tension) in the anterior part and negative (compression) in the posterior region. In contrast, the distal region remains approximately isometric upon knee flexion (see Figure 1). It is accordingly concluded that the sMCL behaves isometric, though large regional differences are observed. The proximal region experiences higher strains. Furthermore, the DIC technique provided valuable insights in the deformation of the sMCL. This technique will therefore be applied to study the impact of knee arthroplasty in the near future. Caption with figure 1: Full field strain distribution in the sMCL's longitudinal direction for specimen in 45° (a) and 90° (b) of knee flexion

INTRODUCTION. Stable fixation of cementless tibial trays remains a challenge due bone density variability within the proximal tibia and the spectrum of loads imposed by different activities. This study presents a novel approach to measuring the interface motion of cementless tibial components during functional loading and tests whether interface motion of cementless tibial trays varies around the implant periphery. METHODS. We developed a method to measure relative displacement of a tibial tray relative to the underlying bone using 3D digital image correlation (DIC) and multi-camera stereo photogrammetry. A clinically successful design of cementless total knee prosthesis (Zimmer Inc, Warsaw, IN) was implanted in 6 fresh cadaveric knees. A black-on-white stochastic pattern was applied to the outer surface of the tibia and the cementless prosthesis. High resolution digital images were prepared of the interface region and divided into 25 × 25 pixel regions of interest (ROI). Stereo images of the same ROI were generated using two cameras angled at 60 degrees using image correlation techniques. All specimens were mounted in a custom-built functional activity simulator and loaded with the forces and moments recorded during three common functional activities (standing from a seated position, walking, and stair descent), as reported in the Orthoload database, scaled by 50% for application to cadaveric bone. Prior to functional testing, each implant-tibia construct was preconditioned with 500 cycles of flexion from 5–100 degrees under a vertical tibial load of 1050 N at a frequency of 0.2 Hz. During loading, image data was acquired simultaneously (±20 μs) from the entire circumference of the tibial interface forming 4 stereo images using 8 cameras spaced at 90 degree intervals (Allied Vision Technologies, Exton, PA) using custom image acquisition software (Mathworks, Natick, MA) (Figure 1). The multiple stereo images were registered using the surface topography of each specimen as measured by laser scanning (FARO Inc., Montreal) (Figure 2). During post-processing, the circumferential tray/tibia interface was divided into 10 zones for subsequent analysis (Figure 3). Interface displacements were measured on a point-to-point basis at approximately 700 sites on each specimen using commercial DIC software (Dantec Dynamics, Skovlunde, Denmark) (Figure 4). RESULTS. The average 3D displacement over 10 circumferential zones of the tray was 83.6±41.5 μm (range: 30.8 to 214.9 μm). The anatomic components of tray migration were 0.4±40.8 μm medially (range: 172 μm lateral to 112 μm medial) and 3.1± 40.6 μm posteriorly (range: 86 μm posteriorly to 61 μm anteriorly). The largest tray displacement was observed in the inferior direction with an average inferior displacement of 37.6±63.8 μm (range: 206 μm inferiorly to 81μm superiorly). The largest displacements were observed posteriorly, with the posteromedial aspect subsiding more the posterolateral aspect. DISCUSSION. The stability of tibial trays cannot be accurately assessed by measuring interface motion at a few fixed peripheral sites. If discrete displacement transducers are used for pre-clinical testing, a set of 4–6 transducers should be placed at sites that vary with the pattern of interface motion of each design and the combination of loads and moments applied during testing