These results show that the developed method allows reproducible in-vitro measurements of the indirect reflect of deformation variations occurring in the CBTPE during knee flexion-extension. We are conscious that the strain gages are designed for being bonded onto the surface of structure but previous study [16, 17] used strain gages into structure. The cancellous bone is not homogenous and anisotropic and the orientation and location of the trabecula are very important to loading transfer. The strain field in structure point is three dimensional. There are three normal strains and three shear strains. In our study we decided consider the measurement of vertical strain. Note that data was dependent of the epoxy resin deformation. The data obtained were the indirect reflect of the cancellous bone load transfer. Nevertheless, the introduction of different structure in the cancellous bone could create a local reinforcement and modify the cancellous bone mechanics. During dynamics of gait, ground reaction force is of primary importance to explain joint loading. In our experimental setting, an open kinematic chain was studied. It would be therefore interesting to reproduce this study in a closed kinematic chain setting to take into account the contribution of ground reaction force, that might affect cancellous bone deformation differently as compared to loading along muscle lines of actions.
Intra-observer reproducibility was satisfactory for both MCV (not exceeding 10%) and CMC (above 0.93). Inter-observer reproducibility indicated that similar measured deformation patterns could be found for all operators at all ME locations (CMC mean: 0.82 to 0.87), although these patterns showed different ranges. It could have been advanced that this range difference could be related to the different velocities applied by the operator to flex the knee joint. We studied the correlation between the mean RMS values and mean and maximal primary motion velocities. The coefficients of determination (r2) between RMS and motion velocity and were 0.32 and 0.39, respectively for mean and maximal velocities. These results do thus not support that velocity was a factor influencing local bone deformation.
For intra-specimen repeatability we chose to analyse the flexion movement. Indeed, no difference between flexion and extension intra-observer reproducibility was observed. Moreover, it seems more logical to express our data according to flexion movement. Indeed, knee joint kinematics and muscular moment arms that are pertinent to interpret deformation data are generally expressed during this movement. The average mean RMS differences (7 to 10%) and the Mean ICC (0.95 to 0.99) showed that maximum variability did not exceed 10% and that a great similarity of the curves was observed. The mean correlation coefficient was ranged from -0.22 and 0.55, indicating that RMS differences were independent of the signal intensity. This implies that the measurement error is constant and does not exceed 10%
The curve pattern of ME6 in all specimens (Figure 5) suggests individual variability of knee deformations. The intensity variability may be due to some discrepancies in gage placement even if this was standardized, to the quality of cancellous bone, especially in elderly people [13, 26]. Indeed, this is approximately 654 (± 304) MPa in young subjects, 829 (± 422) MPa in adults and 613 (± 319) MPa in elderly people . The variability could also be due to individual anatomical and/or kinematical variations (e.g. joint geometry, presence or absence of inconstant ligaments, motion patterns).
Compression tests of the resin cylinders indicated that the average value of the resin Young's modulus was 2.09 (± 0.03) GPa. In comparison to the CBTPE Young's modulus , the resin Young's modulus is 2.4 to 66 times larger than the CBTPE. This means that the epoxy resin deformation is smaller than the cancellous bone's and that the data obtained via the MEs underestimated the real deformation of the cancellous bone. However, this system is satisfactory to answer the main aim of this study, which was developing a method to compare bone deformation variations between two conditions (i.e., before and after osteotomies). For example, Figure 6 showed the CBTPE deformation for ME2 after varum and valgum deviation for another specimen. The data showed that the deformation patterns seems to be in agree with the frontal deviation theory [1, 2] were varum deformity induce a medial shift of the mechanical axis of the lower limb, increasing medial tibial plateau constraint and inversely for valgum deformity. We showed that before 65°, the varum 6° condition decrease CBTPE. After 65° the varum 6° condition increase CBTPE compared to the intact condition. The valgum conditions were not proportional to the degree of frontal deformation. This fact could be due to the modification of muscles and ligaments tensions. But this hypothesis should be still confirmed thanks to confrontation with kinematics and moment arm data. Even if the introduction of a rigid element into the cancellous bone can induce a modification of its mechanic behavior, these preliminary results showed that our methodology allows objective measurement of this problematic.
The tunnel size has been selected after several trials to optimize the contact surface between cancellous bone and ME. The MEs were only introduce on CBTPE and no glue was used. No sliding and movement of the ME in the tunnel were observed. There is currently no other direct method which allows validation of ME output data. To the authors' knowledge, no other 'direct' method is available to record cancellous bone deformations during motion. Only indirect methods exist [12, 13] and these deal with static positions and therefore are not suitable to validate the protocol presented in this study. The latter is the first method which allows to analyze directly in-situ the variations of cancellous bone during a joint movement. The deformation pattern for each individual specimen in some well-defined conditions (e.g., osteotomy) can therefore be compared to each other. Unfortunately, the method does not allow the absolute strain values to be obtained since there is no other direct method is available from the literature for validation.