There are various multi-segment foot models (MFMs) for assessing three-dimensional foot motion in clinical gait analysis [1
]. Each MFM differs not only in the location of the markers on the foot and how the foot segments are defined, but also in the way they are calibrated for the foot’s reference position and coordinate system [2
Majority of MFMs construct the reference frame using three or more markers placed on each segment identified in the static standing trial. After the static calibration, relative angles of the segments while walking are calculated [3
]. This is a general marker-based method of performing motion analysis that builds segments by skin-mounted markers, but it is significantly affected by marker-placement errors among sessions or evaluators [5
]. Some models subtract the static offset values from walking trials [11
]. This can increase repeatability and reproducibility, but the omission of anatomical information is a concern [12
]. Meanwhile, a few models rotate some coordinate systems according to radiographic and/or goniometric measurements during the reference frame construction [13
]. This could be less affected by marker-placement errors and reflect actual bone anatomy. However, this requires subjects to be exposed to radiation, and some measurements are difficult to acquire from radiographic images such as the shank and hindfoot in the transverse plane, and the forefoot in the coronal plane [13
Previous studies using various MFMs for specific foot deformities can be reviewed and compared by researchers and clinicians [1
]; however, these comparisons are limited because it is difficult to understand the special feature of each MFM and its relative differences from other MFMs. Although repeatability studies have already been conducted individually for each model, multiple factors such as the demographic characteristics of the subjects, laboratory environment, operators, statistical analysis, and test intervals were still different, which could have affected their outcomes [5
Di Marco, Rossi [12
] conducted a comparative study on four MFMs. They identified the most repeatable and reproducible model only in the sagittal plane and the kinematic differences between treadmill and over-ground walking without comparing MFMs. Nicholson, Church [17
] also verified, via a comparative study of five MFMs using an amalgamated model, that the MFMs had moderate to low variability as assessed by standard deviations, and that using the same normative data for each model is important when comparing findings between laboratories. However, they did not consider the shank coordinate system that affects the foot kinematics and did not include some methods of applying offset angles, such as radiographic measurements. Consequently, these previous studies have limitations in comparing MFMs simultaneously.
Therefore, this study aimed to compare the kinematics, inter-session repeatability, and inter-evaluator reproducibility among five MFMs of healthy males during walking with all their markers simultaneously in the same experimental conditions with various analytical methods.
In this study, we investigated the kinematic differences, repeatability, and reproducibility of five MFMs that are widely used in the clinical gait analysis and obtained several meaningful results.
According to classifications suggested by Garofalo, Cutti [20
], the inter-session repeatability and inter-evaluator reproducibility of all five MFMs in our study ranged from “very good” to “excellent” in the sagittal plane [7
]. In general gait analysis, the reliability decreased from the sagittal to the coronal, and transverse planes [23
], but the reliability of the coronal plane was lower than that of the transverse plane for HF in the OFM and mRFM models [7
]. Furthermore, the OFM, DFM, and mRFM models, which have a marker-based analysis, showed a lower repeatability of HF in the coronal plane than in the transverse plane compared to MiFM and mSHCG, which use offset angles. There may be more variables such as the influence of the Euler/Cardan angle sequence [25
]; however, we consider that the horizontal variation in the placement of the two posterior heel markers had a major impact on the marker-based analysis of coronal HF motion. Thus, clinicians or researchers using only marker-based analysis should consider using a calcaneal marker placement device [26
] or establish a more precise criterion for locating these markers and sticking them carefully.
McGinley, Baker [27
] stated in their review that in common clinical situations, an error of 2° or less was highly likely to be acceptable. Errors between 2° and 5° were also likely to be regarded as reasonable but may require consideration in data interpretation. Schwartz, Trost [21
] introduced experimental errors in the lower extremities (excluding the multi-segmented foot), and the mean σeval
ranged from 1.2 to 5.3°. In a study that applied σ to MFM, Deschamps, Staes [8
] reported that mean σsess
ranged between 0.9° and 5.0°, while the mean σeval
ranged between 2.8° and 7.6°. Saraswat, MacWilliams [14
] verified that mean σsess
were less than 6.0°. In our study, MiFM and mSHCG showed lower σ than other models in motion analysis with offset angles, but the σtrial
, and σeval
of all MFMs did not exceed 3.5°. This indicated that the highest σ in our study was lower than those of previous studies. Therefore, we believe that rotating the segments by the offset angles obtained from radiographic and/or goniometric measurements increased reliability, and consequently avoided the effects of marker-placement errors. However, any model used in our study would be clinically acceptable.
MiFM and mSHCG showed greater HF dorsiflexion and FF plantarflexion compared to other models because they reflected the pitch angles of the calcaneus and FF [13
]. Additionally, we verified a close affinity in HF for all planes between OFM and DFM, which corresponds with the findings of Nicholson, Church [17
]. However, the CMCs for DFM of HFtrans
were lower than those for OFM. This suggests that the DFM, by using markers on the medial/lateral malleolus to coordinate the HF segment, was more variable in the transverse plane than the OFM, which applies markers on the medial/lateral calcaneus.
Although most MFMs showed no statistical differences in HF varus/valgus SPM curves except for the toe-off of mRFM-mSHCG, DFM and mSHCG showed significantly increased ROM. In the HF external/internal rotation, there were also similarities in kinematics and ROM in DFM, mRFM, and OFM; however, MiFM and mSHCG showed inconsistent kinematics and decreased ROM compared to the other models. In addition, the point of peak angle showed large deviations in some motions and significant differences in FF. We think that these dissimilarities were not due to the offset angle but to the different local coordinate system and marker placement for each MFM. In particular, soft tissue artifacts that occur differently in each model due to the marker placement discrepancy even within the same segment also had a critical influence [28
]. In previous studies, the wand marker, which was used for mSHCG, reflected only 40–70% of the actual axial hip rotation when attached to the lateral thigh [30
]. Similarly, the three markers on the lateral shank, which were used for DFM, can rotate themselves by moving with the calf muscle, creating excessive rotation in the proximal segments [17
]. In other words, the sensitivity to motion was different for each MFM due to the influence of differences in the segment coordination and the marker type and location. Hence, these factors must be considered when comparing clinical studies using different MFMs.
We obtained meaningful results by comparing the kinematic characteristics of the five models, but it was impossible to find a model that accurately depicts actual foot motions. To compare MFMs with actual foot movements, the influence of STA must be considered. Although there was a study that measured rear, mid, and forefoot kinematics and ROM through an invasive in vivo study using bone pins [31
], it could not be compared with our results because the subjects and experimental environments were different. In addition, valuable studies have been conducted to quantify the STA between the skin-mounted marker and the bone to identify the location most affected by STA [32
]. They reported that the medial malleolus [32
], lateral malleolus [34
], navicular [32
], medial calcaneus [32
], lateral calcaneus [33
], and posterior aspect of the proximal calcaneus [34
] were significantly affected by STA during maximum plantarflexion. In particular, Schallig, Streekstra [34
] reported that, in a study with a computed tomography scan, RFM that utilized the posterior aspect of the proximal calcaneus marker as a tracking marker was significantly affected by STA, compared to OFM, which used the proximal calcaneal marker only in the anatomical coordinate system. In summary, although we could not compare the bone movement in five models, further studies are needed to find a model that most accurately reflects actual foot movement.
Some limitations should be considered when appreciating these results. First, only young adult men were analysed in this study. The biomechanics of the elderly, children, and females may differ from those of young males. Second, although the number of subjects in this study was small, it is similar to that of other studies [7
]. Further research is required to investigate the implications of the findings to a wider population. Finally, we applied some slight modifications of marker placements to mRFM, OFM, and mSHCG for convenience. This could affect the FF kinematics of mRFM and those of HF of OFM and mSHCG.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.