New anatomical reference systems for the bones of the foot and ankle complex: definitions and exploitation on clinical conditions

In human joints in general, and more specifically in the foot and ankle complex, a careful definition of relevant anatomical reference systems (ARS) is essential for the quantification of the absolute and relative position and orientation of the bones, thus allowing the description of the foot posture and motion in healthy and pathological conditions. Robust definitions of ARS must guarantee high inter- and intra-operator repeatability, consistency over follow-up assessments and among similar subjects, clinical relevance and interpretability, and a good alignment with joint axes of motion to optimize the kinematic description [1‐3].

Early definitions of foot-ankle related ARS were provided for gait analysis via stereophotogrammetric systems and were based on external markers, meant to represent corresponding anatomical landmarks [4, 5]. However, due to the complexity in the identification by palpation of externally accessible landmarks for all the foot bones [6], ARS were defined mostly for multi-bone segments [7] [8]. Furthermore, it was impossible to define the ARS for the talus due to the direct inaccessibility of its landmarks [9, 10]. This marker-based approach provided insight in the general posture of the foot during functional tasks but did not allow its complete kinematic description [6]. Also, identification of anatomical landmarks by external palpation resulted in a low ARS repeatability [11].

The development and dissemination of new biomedical imaging techniques, such as Weight Bearing CT (WBCT) scanners based on the cone-beam technology, is providing easier access to a complete 3D representation of the foot and ankle bones, opening the way to new approaches in ARS definition [12]. In the literature, alternative approaches are based on landmarks [13, 14], principal inertial axes [15, 16], morphological fitting of geometrical features [17, 18] or a combination of the previous techniques [19, 20]. Anatomical landmarks can provide good clinical interpretation [21], but they are limited by the operator-dependent accuracy and repeatability [22]. Principal inertial axes can be determined automatically and thus do not depend on the operator [23]; however, the resulting ARS may be not directly associable with anatomical planes [21] nor optimal for kinematic description [1], and axis orientation is not uniquely defined [24]. Another possible approach consists in defining clinically interpretable ARS for one bone (typically the tibia) and assuming the same orientation of the ARS for all the foot bones in a given scan, for instance in ankle neutral configuration [25]. This is a simple and fast method, particularly suitable to describe the relative kinematics, but unable to provide a clinical description of the foot posture [26]. Finally, the morphological approach allows the identification of functionally relevant features that, despite being still sensitive to the operator [1], are more repeatable with respect to punctual landmarks thanks to the wider dimension of the fitting region. Furthermore, the axes identified through morphological fitting may provide a good approximation of the joint axes of motion, as shown for the tibiotalar and subtalar joints [17, 27]. Despite its potential, this latter approach is currently applied to the talus bone only and its application has not been fully exploited.

The variety and the number of the available approaches, also reflected in the number of different clinical measures to assess foot posture [28], suggest that a robust definition of ARS for all the bones of the foot and ankle is still lacking, making it difficult to perform a thorough comparison of different data from the literature.

The goal of the present study is to propose a new definition of ARS for the foot and ankle bones using a morphological approach mainly based on the shape of the articular surfaces. More specifically, we targeted ARS definitions that are: (a) highly repeatable; (b) consistent among individuals; (c) clinically interpretable; (d) optimized for the kinematic description. An algorithm is also provided that, given the appropriate geometrical features fitted on a 3D representation of the bones model, computes ARS for each bone in the foot and ankle complex.

The overall foot and ankle ARS are depicted in Fig. 9. To help their observation, ARS are also shown for the different groups of bones composing the foot and ankle complex: the hindfoot, the midfoot, and the forefoot (Fig. 10).

The aim of this work was to define robust anatomical reference systems for each bone of the human foot and ankle complex. To this end, a procedure based mainly on morphological fitting of simple geometrical features on the articular surfaces is presented, also implemented in a complementary software code provided with the present paper.

Our analysis shows that the proposed definition is highly repeatable for all these bones, both inter- and intra- operator. Moreover, the identification of ARS is highly consistent among individuals, providing a solid representation of the three-dimensional posture of the foot and ankle complex. This representation is also clinically relevant, as shown for instance by the quantification of the curvature of the foot arches in normal and flat feet, whose values compared well with those observed in standard clinical measurements from planar radiographs. With respect to these, the present definitions clearly allow geometrical representations of position and orientation in three dimensions, thus extending standard measurements. Finally, the presented ARS are optimized for joint motion description, as proven by the alignment between mean helical axes and specific reference axes, particularly in the talar and calcaneal reference system. This suggests that the proposed ARS may also describe motion of the foot joints well, minimizing the apparent coupling among different motion components.

Although the proposed ARS may provide valuable clinical information, they are not meant to substitute standard radiographic inspection, which due to the simplicity in the setup, the diffusion and the established use will remain the gold standard for traditional clinical assessment of foot pathologies. Rather, the proposed ARS are intended to maximize the opportunities presented by the emerging techniques such as the WBCT, which makes it possible to investigate overall foot and ankle posture in three dimensions, opening the way to new 3D measures [33] and providing a solid framework for the investigation of foot motion.

While similar approaches have been previously proposed, particularly for the tibia and talus [12], a systematic use of morphological fitting to define ARS in the foot has not been previously reported in the literature. In comparison with other possible approaches, morphological fitting shows several advantages. With respect to virtual identification of anatomical landmarks, the operator must identify broader surfaces. Thus, the uncertainties introduced by the operator are smoothed by the dimension of the selected areas and by the fitting process. As a result, the approach presented is more repeatable and consistent than others based on virtual palpation of anatomical landmarks [21]. Also, the present ARS axes are optimally oriented with respect to axes of joint motion [1].

When compared with the fully automatic methods based only on principal axes of inertia [29, 33] for the definition of reference systems, the approach here presented has the advantage to be more clinically relevant in genreal [21], still showing a very high repeatability. In terms of consistency however, the presented ARS perform better, since the principal axes of inertia are indeed quite sensitive to bone geometry for axes different from the main one, particularly in case of long-cylindrical objects (such as the metatarsal bones), and in general in case of stocky objects such as cuneiforms or cuboid.

The proposed approach has limitations. Its application requires a full three-dimensional model for each bone of the foot and ankle complex, with the exception of the tibia and the fibula for which the distal portions are sufficient. The number of subjects analyzed needs to be increased to further test the consistency of the results. The manual identification of the necessary geometrical features and the calculation of all the ARS require some time. For this reason, we provide the code for the automatic computation of the ARS in the supplementary material, once the parameters of the fitted geometrical features have been defined by an operator. With this, the definition of the ARS for the whole foot and ankle complex takes roughly 30 min. Also, despite the excellent repeatability, the approach still depends to a certain extent on the operator. For both these reasons, future activities will aim at the full automatization of the process.

Finally, the present work represents the preliminary step toward an ongoing investigation of the foot kinematics under loads. Future activities will exploit the proposed ARS to describe the variation in the foot posture with ankle dorsi-plantar flexion and foot pronosupination thoroughly.

According to a morphological approach, careful definitions of anatomical reference frames for all bones in the foot and ankle complex are proposed and exploited in a number of feet. These definitions proved to be robust, being found highly repeatable and consistent among individuals within the same population. Also, the proposed ARS are clinically interpretable, thus providing quantitative measures to be used in clinical studies. Finally, these definitions were shown to be suited for the description of joint kinematics. Therefore, the present approach provides a solid base for the 3-dimensional description of posture and motion of the foot and ankle skeletal structures.