Full length articleAnatomical masking of pressure footprints based on the Oxford Foot Model: validation and clinical relevance
Introduction
Plantar pressure analysis is an effective tool for measuring foot function [1], [2]. When used clinically alongside lower-limb kinematic and kinetic data, the results obtained can contribute to management planning and treatment decision-making.
A common method used to analyse plantar pressure parameters involves masking the footprint into regions of interest (ROIs). This can provide more descriptive and clinically relevant information than when examining the foot as a whole [2], [3].
Manual masking involves visual examination of the footprint, and selection of sub-areas based on subjective identification of areas believed to correspond to anatomical structures of interest. The accuracy of this process depends on the spatial and pressure resolution of the platform, the anatomical knowledge of the clinician, and the clarity with which each anatomical structure may be identified ([4], [5]). Manual masking may be helpful or unavoidable under some particular conditions, as in the presence of foot deformity, but the lack of automation renders its repeatability questionable, and thus is rarely used.
Automated techniques are based on masking algorithms which may be grouped into two main categories, – geometry-based (GM) and anatomy-based (AM) masking. GM methods are the most widely used both in research and clinical settings and involve the pressure footprint being divided up based on geometric features. These methods are recognised as being reliable and repeatable. There are different approaches to GM masking [3]. The simplest methods rely only on footprint geometry, without reference to measured pressure. Typically, the footprint is divided into pre-defined subareas based on medial and lateral tangents of the footprint, a bisecting longitudinal line, and lines drawn perpendicular to the bisecting line in correspondence with specific percentages of foot length [6]. More complex GM methods also exploit pressure gradients and pressure distribution within the footprint map so as to better refine the selection and identify anatomical structures such as metatarsal heads, the first metatarsal-phalangeal joint and minor toes [7], [8], [9]. The major limitation of GM methods is the lack of ability to automatically mask incomplete or significantly altered footprints.
To overcome this limitation, automated and objective footprint masking based on anatomical structures of the foot (AM) was developed [10]. This method relies on the integration of a 3D motion capture system and associated tracking markers, as well as a plantar pressure measurement device. The advantage of AM methods over GM methods is the use of actual anatomical landmarks rather than arbitrary divisions of the foot, while maintaining automation and repeatability. Further, with respect to the previously mentioned manual anatomical masking, the main advantage of AM consists of improved repeatability and greatly reduced human error due to the subjective identification of these landmarks [11], [12], [13].
Clinical implementation of AM has been described in various pathological populations with encouraging results, including: i) subtalar coalition [14], in conjunction with the IOR Foot Model [15]; ii) cerebral palsy [11], [12], [13] through use of the Oxford Foot Model OFM, [16], and iii) diabetes [17]. In the field of gait biomechanics research, MacWilliams et al. [18] reported on the validation of a nine-segment foot model. Their AM based on this model consisted of 6 ROIs namely rearfoot, medial and lateral forefoot (also including midfoot), hallux, central toes and lateral toes. More recently, a study based on the updated IOR foot model [19] showed the potential of AM not only to improve identification of ROIs, but also with respect to correlation analysis between kinematics of specific foot segments and loading parameters of corresponding plantar regions [20].
Despite these interesting clinical and research implementations, a thorough validation of AM has not yet been reported. Giacomozzi et al. [10] performed the preliminary “technical” validation of the whole procedure, showing that: the instrumentation used had adequate accuracy; the synchronization and re-alignment of the measuring systems were feasible; the matching between pressure footprint and foot anatomy was acceptable; the proposed anatomy-based regionalisation was reasonable and applicable to healthy volunteers. However, an extensive validation using pathological footprints is still lacking. The present study thus aims to validate the AM approach to footprint masking by comparing its performance with that of the most reliable, currently accepted approach. Specifically, 5 ROIs based on the OFM, designed to address the specific needs of paediatric clubfoot (CF) are presented. GM is currently recognised as a standard, reliable and objective method for identifying ROIs during barefoot level walking. Therefore, to validate AM performance, this method should be at least as reliable as GM. It is thus hypothesised that the coefficient of variation (CV) within the repeated trials of each individual subject will not be statistically different when compared to GM. The clinical relevance of AM is investigated by comparing the ability of the AM method to detect meaningful differences between healthy and CF matched footprints to the performance of a comparable GM method.
Section snippets
Recruited population
20 healthy volunteers (HG: age 11.5 ± 2.8 years, BMI 18.1 ± 3.1 kg/m2) and 20 patients with treated clubfoot (CF: age 11.0 ± 3.3 years, BMI 19.5 ± 4.0 kg/m2) were assessed at the Oxford Gait Laboratory (OGL) after their parents/guardians had signed informed consent. The study was approved by the local ethics committee. HG volunteers were selected and the group was matched to CF patients in terms of age and BMI. Exclusion criteria for HG were presence of musculo-skeletal problems or pain, abnormal walking
Results
40 feet, for a total of 160 footprints, were analysed for the 20 healthy volunteers (HG) and 39 for the CF group (156 footprints in total). No statistical differences were found between HG and CF demographics and anthropometrics, or within the CF subgroups, except for the flatfoot group (F-CF) (Table 1).
Discussion
The goal of anatomy-based masking applied to plantar pressure footprints is to associate regional loading, or a change in loading, to corresponding anatomy and/or function of the foot, ankle or even lower leg structures. AM methodology may therefore represent a valuable tool not only for dealing with severely altered footprints, where definition of ROIs on the basis of geometrical criteria cannot realistically be performed, but also to define anatomically accurate, more robust and standardized
Conflict of interest
None.
References (24)
A review of the utilization of baropodometry in postural assessment
J. Bodyw. Mov. Ther.
(2014)- et al.
A cross-sectional study of age-related changes in plantar pressure distribution between 4 and 7 years: a comparison of regional and pixel-level analysis
Gait Posture
(2014) - et al.
Age-related differences in foot structure and function
Gait Posture
(2007) - et al.
Clinical determinants of plantar forces and pressures during walking in older people
Gait Posture
(2006) - et al.
Between-day reliability of repeated plantar pressure distribution measurements in a normal population
Gait Posture
(2008) - et al.
Assessment of sub-division of plantar pressure measurement in children
Gait Posture
(2005) - et al.
An anatomically based protocol for the description of foot segment kinematics during gait
Clin. Biomech.
(1999) - et al.
Repeatability of a model for measuring multi-segment foot kinematics in children
Gait Posture
(2006) - et al.
Integrated kinematics-kinetics-plantar pressure data analysis: a useful tool for characterizing diabetic foot biomechanics
Gait Posture
(2012) - et al.
Foot kinematics and kinetics during adolescent gait
Gait Posture
(2003)