Elsevier

Gait & Posture

Volume 39, Issue 1, January 2014, Pages 577-582
Gait & Posture

Foot roll-over evaluation based on 3D dynamic foot scan

https://doi.org/10.1016/j.gaitpost.2013.09.014Get rights and content

Highlights

  • 10 subjects have been assessed during walking with a dynamic foot scanner.

  • A method for foot roll-over evaluation was presented and applied on collected data.

  • Reliability was tested on some computed variables.

  • The potential relevance of this method was discussed regarding pathological gait.

Abstract

Foot roll-over is commonly analyzed to evaluate gait pathologies. The current study utilized a dynamic foot scanner (DFS) to analyze foot roll-over. The right feet of ten healthy subjects were assessed during gait trials with a DFS system integrated into a walkway. A foot sole picture was computed by vertically projecting points from the 3D foot shape which were lower than a threshold height of 15 mm. A ‘height’ value of these projected points was determined; corresponding to the initial vertical coordinates prior to projection. Similar to pedobarographic analysis, the foot sole picture was segmented into anatomical regions of interest (ROIs) to process mean height (average of height data by ROI) and projected surface (area of the projected foot sole by ROI). Results showed that these variables evolved differently to plantar pressure data previously reported in the literature, mainly due to the specificity of each physical quantity (millimeters vs Pascals). Compared to plantar pressure data arising from surface contact by the foot, the current method takes into account the whole plantar aspect of the foot, including the parts that do not make contact with the support surface. The current approach using height data could contribute to a better understanding of specific aspects of foot motion during walking, such as plantar arch height and the windlass mechanism. Results of this study show the underlying method is reliable. Further investigation is required to validate the DFS measurements within a clinical context, prior to implementation into clinical practice.

Introduction

Foot roll-over is often evaluated using plantar pressure plates, insole sensors or single transducer systems [1]. Independent of the system, output data are often similar. A footprint image is obtained for each frame of the stance phase, with a pressure intensity value for each pixel available from the image. Plantar pressure data are generally represented by a peak pressure picture. This gives an assessment of foot contact in one single image. Peak pressure images are also used as to represent pressure data in anatomical regions of interest (ROIs) [1]. Conversely, dynamic analysis is necessary to characterize temporal (e.g., contact duration, instant of contact), spatial (e.g., contact area) and intensity (e.g., mean pressure, impulse) properties in each ROI.

Recently, another approach has been proposed by Mochimaru and Kouchi [2] to evaluate roll-over in one image (representation). A pattern was projected onto the foot sole during gait using a projector integrated into a glass walkway. The reflected pattern was captured by the camera to obtain dynamic representation of the 3D foot sole. These 3D data are similar to plantar pressure data in that it represents time series data consisting of a 2D shape with a third dimension [2]. The “lowest height data” (LHD) picture was computed summarizing foot roll-over in one foot sole picture, similar to a peak pressure image obtained from pedobarographic measurements. As illustrated on Fig. 1, the LHD matched the minimum vertical coordinate (z) recorded for specific horizontal coordinates (x,y) during the complete stance phase. Consequently, LHDs on the rearfoot, the midfoot and the forefoot are mainly recorded during touchdown, flat foot contact and heel-off, respectively. Using height data, the main difference to plantar pressure was the ability to evaluate some foot sole areas without any physical contact with the ground during foot roll-over (e.g., medial part of midfoot sole).

Nevertheless, even though this approach was the first to present information about foot roll-over based on 3D data, two main limitations were present. First, use of only one camera was insufficient to record the complete 3D foot sole during each instant of the step (i.e., borders of plantar sole shape), introducing some possible errors during analysis (e.g., foot orientation, contact area). Secondly, LHDs summarize foot roll-over in only one footprint picture. A frame-by-frame output is required for dynamic analysis.

Recent studies have presented initial data using dynamic foot scanners (DFS). The complete 3D foot shape is able to be tracked during gait using multi-cameras systems [3], [4], [5], [6], [7]. Using this kind of DFS technology, the goals of this paper were to extend the LHDs method in a dynamic situation, and to identify new information or possible applications made possible by this dynamic approach.

Section snippets

Experimental set-up

The right feet of ten healthy subjects (26.2 ± 4.1 years, 1.73 ± 0.07 m, 70.5 ± 6.9 kg) was recorded during five gait trials at self-selected speed on a DFS integrated into a walkway. The DFS (Dynamic Foot Morphology, Lion Systems S.A., Foetz, Luxembourg) was composed of a box covered by a glass plate and three time-of-flight cameras (40 Hz, 176 × 144 pixels) for 3D data acquisition [4] with accuracy close to 0.5 mm [8]. The cameras recorded the foot during gait from the medial, lateral and plantar (with

Results

Fig. 4 shows the subjects’ mean curves for each ROI mean height and projected surface, as well as the corresponding ICC values. Details of all ICC values are given in Appendix 1. For each ROI, the first and the last frames with a height are lower than 15 mm are displayed (dotted lines in Fig. 4). The other frames were computed for mean height and projected surface respectively equal to 15 mm and 0% of full surface. The mean height of the medial rearfoot and the three forefoot ROIs evolved

Discussion

The current results display differences compared to traditional plantar pressure data [10], [11], [12], [13]. The differences between height data and plantar pressure data are explained by the specificity of each physical quantity. Height data and pressure are not systematically linked to each other: different pressure quantities could be recorded with the same height data, and vice versa. For this reason, it was hypothesized that height data would provide new information about foot roll-over

Summary

The current approach was proposed to extend the LHD method using a DFS system. A foot sole picture was produced from by vertically projecting points from the 3D foot shape with a height threshold of less than 15 mm. A height value for these projected points was assigned corresponding to their initial z coordinates. The mean height and projected surface variables were computed for seven ROIs from foot sole picture. ICCs revealed that intra-subject variability for these variables was

Acknowledgements

The authors would also like to acknowledge Claude Seiler for his data processing assistance during the study. The present project is supported by the National Research Fund, Luxembourg.

Conflict of interest statement: The authors have nothing to disclose.

References (19)

There are more references available in the full text version of this article.

Cited by (11)

  • Stature estimation from various three-dimensional anthropometric foot measurements of Taiwanese female population

    2022, Legal Medicine
    Citation Excerpt :

    Foot length (FL) is defined as the distance along the foot axis from the end of the heel (Pternion) to the tip of the longest toe [7,13,24]. The foot axis is referred to as a straight line from the heel center to the forefoot center (middle of the axis connecting the 1st and 5th metatarsal heads) [28]. The ball of foot length (BFL) and outside ball of foot length (OBFL) are the distances from the end of the heel to the 1st and 5th metatarsal point protrusions [13,19], respectively.

  • Foot Type Biomechanics

    2022, Foot and Ankle Biomechanics
  • 3D Surface Scanning of the Foot and Ankle

    2022, Foot and Ankle Biomechanics
  • A state-of-the-art review of foot pressure

    2020, Foot and Ankle Surgery
    Citation Excerpt :

    Table 2 summarizes the advantages and disadvantages of the plantar platform and in-shoe system. The plantar platform system and the in-shoe system require different kinds of force sensors such as the semiconductor strain gauge transducers [10], critical light deflection Dynamic Foot Morphology, Lion System S.A., Foetz, Luxembourg [47], capacitive strain gauge [48], capacitive sensors – emed® platform systems and Pedar® in-shoe system (Novel, Germany) [20,30], resistive sensors (also known as force-sensing resistor, FSR) – MatScan® platform system and F-Scan® in-shoe system (TekScan, Boston, MA) [17,22] and force sensor with light deflection Biokinetics dynamic optical pedobarograph (Biokinetics Inc., Bethesda, USA) [49]. Each system can have different number sensors to provide an electrical signal output proportional to the measured foot pressure; usually 1.4 [22] to 4 [14,36] sensors per cm2 of the plantar platform and 3 to 15 sensors locations with 0.391 cm2/sensors per location for the in-shoe system [30].

  • Dynamic measurement of surface strain distribution on the foot during walking

    2017, Journal of the Mechanical Behavior of Biomedical Materials
    Citation Excerpt :

    To the best of our knowledge, this is the first attempt to quantify temporal changes in the foot surface strain distribution during walking. The 3D deformation in foot shape during walking has previously been captured using video cameras and optical scanners (e.g. Kouchi et al., 2009; Jezersek and Mozina, 2009; Kimura et al., 2011; Novak et al., 2014; Samson et al., 2014; Van den Herrewgen et al., 2014), but the change in skin strain during walking has not been quantified because 3D scanners cannot track the change in position of a certain point on the foot surface during motion. Although motion capture systems can overcome this problem and track the movements of markers, the number of markers that can be placed on the foot surface is limited, prohibiting these systems from capturing skin strain distribution in sufficient detail.

View all citing articles on Scopus
View full text