Comparison with existing literature
Full RoM of MTP1 in sagittal plane of 40.0° is comparable with results on cadaver feet [
13] and existing multi-segment foot models [
2,
3,
5,
18], also the timing of plantar flexion and dorsal flexion are visually comparable with existing data. Full RoM in the transverse plane of MTP1 is comparable to results from cadaver experiments [
13]. However, the timing in the cadaver experiments is different, for instance in the GM-model no abduction was seen at the end of stance phase. This difference in timing might be caused by the timing of the external forces driving the cadaver experiments. Comparing with other multi-segment foot models that include MTP1 no consistency has been found between the various multi-segment foot models [
2,
3,
5,
18]. As in other models, RoM for MTP1 in transverse plane in the GM-model also shows a low COV25, the reason for this could be a large inter subject variation or a DoF that is hard to measure.
Typically, the RoM of MTP joints 2–5 are not included separately in multi-segment foot models. However, MacWilliams et al. [
5] used a model that separated the phalanges in medial (second and third) and lateral (fourth and fifth) phalanges. Compared to our data, MacWilliams et al. [
5] showed larger RoM for the lateral phalanges in the sagittal plane. However, the shape of the curves was similar to our findings. This could be caused by the differences between the participant groups, the current study involved adults, while MacWilliams et al. were measuring on adolescents (12.49 ± 2.6 years), Nigg et al. showed a decreasing flexibility with age in the foot [
20]. For the transverse plane our results are comparable with the results of MacWilliams et al. [
5] for shape and magnitude. Although the GM-model shows a large inter-subject variation for MTP2-5, comparison with 1 other experiment [
5] shows a moderate comparison.
Motion for individual TMT-joints has not been reported using skin mounted markers. However, MacWilliams et al. [
5] did report motion between medial metatarsals and tarsal bones (mean RoM in sagittal plane 15°, in transverse plane 7°) and lateral metatarsals and tarsal bones (mean sagittal RoM 10°, in transverse plane 4°). Results of the GM-model showed a much lower RoM in sagittal plane for TMT I-V. However, transverse plane motion for TMT IV (8°) and V (7°) is higher in the GM-model compared to the MacWilliams model. The Ghent Foot Model [
3] divided the TMT motion in TMT I and a combined motion for TMT II-TMT V. In this model, TMT I motion is defined with respect to all tarsal bones in contrast to the GM-model in which it is only linked to the medial cuneiform, however motion pattern and magnitude in transverse and sagittal plane differs compared to our results. In the sagittal plane an opposite motion is noticed for TMT I and TMT II – TMT V in the Ghent Foot Model, this large difference seems to be an effect of the rigid body assumption to combine Metatarsal 2–5 and all tarsals. Alternatively, Leardini et al. [
9] measured combined tarsal and metatarsal bones, transverse and sagittal plane motion was in the same range as TMT IV and TMT V of the GM-model. However, the pattern is not comparable, this could be caused by the combination of all tarsals and all metatarsals into two segments by Leardini et al. [
9]. Since the GM-model shows different patterns for TMTI-TMTV, the summation of these different patterns into one motion of one joint that describes all these separate motions leads to a motion that cannot be compared.
Previous kinematic foot models have taken all tarsal bones as one segment, therefore the midfoot kinematics of the GM model can only be compared with cadaver [
13] and bone-pin studies [
19]. Compared to cadaver studies, the RoM of medial cuneonavicular joint in the sagittal plane was lower in the GM-model during start of stance phase but was similar at the end of the stance phase. No consistent pattern was observed for this joint in the bone-pin study of Lundgren et al. [
19]. In the transverse plane the cadaver and bone-pin studies measured opposite motion. Our results are closer to the bone-pin study [
19]. The RoM of the other cuneonavicular joints (central and lateral cuneonavicular joint) has only been measured in cadaver experiments, for both joints sagittal joint motion was lower in the GM-model, however standard deviation in the cadaver experiments was very high (5.1° ± 9.8° and 3.6 ± 14.3°), where the consistency for the GM-model was moderate to high (0.85 and 0.34). The RoM in the transverse plane of central cuneonavicular joints was of the same magnitude and has a similar pattern of motion for the GM-model and the cadaver experiments; however the cadaver experiment showed a large variance (2.2° ± 5.4°).
The calculated RoM of talonavicular joint of the GM-model is partly comparable with previous bone pin data [
19], with a late stance plantar flexion, mid stance eversion. However, transverse plane motion of the GM-model shows a different pattern, comparable to the cadaver experiments [
13]. Calcaneocuboid joint motion in frontal and sagittal plane followed the same pattern as the cadaver experiments. However transverse plane motion was of a different pattern in the cadaver experiment and bone pin studies. The RoM of calcaneocuboid joint in GM-model was similar in pattern and magnitude as reported by the MacWilliams model [
5], which is currently the only kinematic model that measures this joint. Other multi-segment foot models only measured complete Transverse tarsal (or Chopart’s) joint. We did not combine the calcanocuboid and the talonavicular joint RoM of the GM-model into a combined joint, since these two separate joints show a different pattern.
To the authors’ knowledge dynamic RoM of inter-phalangeal joints was not measured before, this might be because of the high resolution needed to accurately measure the motion of small bones.
All in all, no consensus was found in existing literature for most of the joints. However, the novel GM-model showed RoM’s within the variations reported across existing literature. The source of the difference within existing literature and when compared with the GM-model can be in different segment and joint definition and marker placement. Furthermore, there could be compensation in directions in which no motion was allowed, most joints have 2 DoF and no motion is allowed for eversion and inversion, while cadaver and bone-pin studies have shown physical motion in all three planes. Therefore physical motion in a non-modelled direction, can be seen by the model as a motion in a direction that is modelled, this leads to a wrong prediction of the motion.
Another source of differences between existing literature and the results of the GM-model is the different angle definition. The GM-model used joint angles in the local orientation of that joint. While cadaver experiments and bone pin studies do not model joints, but consider absolute differences between the orientations of two bones with respect to the axis of the measurement system. Therefore more distal joints will have a larger deviation, since their orientation is influenced by proximal joints.
Limitations
Limitations of the study are: 1) the small number of participants, 2) the comparison with literature, rather than with subject specific golden standard data, 3) no separate analysis on the effect of scaling and 4) the lack of an inter- and intra-subject variations analysis.
One of the limitations of the current study is that data from six out of ten participants was used. The exclusion of three of the participants was caused by an incorrect marker tracking. This is probably caused by the number of cameras used in the experiment, since these three participants were measured at the Maastricht site, with an eight camera-set up. Due to this small number of cameras it is difficult to position the cameras. On the one hand they should be close enough to capture the small markers on the foot. On the other hand the cameras should have sufficient distance to the foot to avoid reflection of the foot and to increase the field of view. The bandwidth of this positioning is small and is a source error, which can be reduced by increasing the number of cameras or the resolution of the cameras.
The measurement of foot and ankle kinematics currently does not have a gold standard to measure bone motion in a non-invasive, three dimensional, dynamic situation. We have compared results of the GM-model with available data sources. However, a better validation would be possible when novel techniques are further developed, e.g. dynamic MRI and 3D fluoroscopy [
21].
Another limitation is that no experiments have been performed to the effect of the scaling. To decrease the error caused by marker misplacement, it has been chosen to use an extra surface scan to scale the model. However, this scaling is based on 16 points on the foot and ankle, in contrast to the 30 markers used to trace the dynamic system. A sensitivity analysis should be performed to calculate the effect of the scaling.
Before using this model for clinical research it is necessary to study intra- and inter-session variations [
22]. This study only focussed on the development of the model and the validation of its output. Intra session variations are expected to be small, due to the use of the kinematic analysis of an over-determinate system [
17].
The rhythms coupled the motion of various joints; therefore the total number of DoF is decreased. Nevertheless, this model shows the ability of driving all segments. Due to the set-up of the model, it is possible to adapt the model to the needs of a specific research question. It is also possible to model the biomechanical aspects of a specific pathology by adapting the parameters of the rhythms or change or remove and add certain rhythms if, for example, a joint is fixed by an arthrodesis. Since the position of all bones can be computed with this kinematic model, this model will be included in a novel musculoskeletal 26 segment foot model. Since muscles are attached to all bones of the foot, motion within the foot leads to different work lines for muscles and therefore different force characteristics.