Identifying gait asymmetry using gyroscopes—A cross-correlation and Normalized Symmetry Index approach

Abstract

Injury to a lower limb may disrupt natural walking and cause asymmetrical gait, therefore assessing the gait asymmetry has become one of the important procedures in gait analysis. This paper proposes the use of wireless gyroscopes as a new instrument to determine gait asymmetry. It also introduces two novel approaches: normalized cross-correlations (Cc_norm) and Normalized Symmetry Index (SI_norm). Cc_norm evaluates the waveform patterns generated by the lower limb in each gait cycle. SI_norm provides indications on the timing and magnitude of the bilateral differences between the limbs while addressing the drawbacks of the conventional methods. One-way ANOVA test reveals that Cc_norm can be considered as single value indicator that determines the gait asymmetry (p<0.01). The experiment results showed that SI_norm in asymmetrical gait were different from normal gait. SI_norm in asymmetrical gait were found to be approximately 20% greater than SI_norm in normal gait during pre-swing and initial swing.

Introduction

Gait in healthy individual is fairly symmetrical with minor deviations. In pathologic gait, significant differences in kinematic and kinetic parameters can be observed in left and right human lower extremity. These differences include differences in stride time, stance time and swing time (Dewar and Judge, 1980), differences in ground reaction force profiles (Kim and Eng, 2003, Balasubramanian et al., 2007, Crowe et al., 1996, Morita et al., 1995) and differences in the range of motion (Haddad et al., 2006; Patterson et al., 2010; Andres and Stimmel, 1990, Shorter et al., 2008). Hence, gait asymmetry is commonly used as a strong indicator in rehabilitation and clinical settings to identify and track various clinical related problems in patients with cerebral palsy (Seeger et al., 1981), stroke (Jorgensen et al., 2000, Brandstater et al., 1983, Griffin et al., 1995, Balasubramanian et al., 2007, Patterson et al., 2008), and amputation (Prince et al., 1992, Andres and Stimmel, 1990, Mattes et al., 2000).

Most common approaches used to define gait asymmetry are the Symmetry Index (SI), Symmetry Ratio (SR) and the statistical approaches (Sadeghi et al., 2000). Despite of their implementations in various clinical applications (Becker et al., 1995, Kim and Eng, 2003, Balasubramanian et al., 2007, Andres and Stimmel, 1990), SI and SR are susceptible to artificial inflation (Herzog et al., 1989). They also require a reference value which is different from case to case (Zifchock et al., 2008). Moreover, SI and SR fail to analyze the motion in one complete gait cycle (Shorter et al., 2008; Crenshaw and Richards, 2006). On the other hand, statistical approaches are also used to identify gait asymmetry. Among them are the principal component analysis (Sadeghi, 2003), regions of deviation analysis (Shorter et al., 2008) and paired t-test (Allard et al., 1996). However, these methods require additional subjects and experiments and may need normative data from able-bodied subjects as a reference.

So far, the research directions were concentrated on the ground reaction force profiles (Kim and Eng, 2003, Balasubramanian et al., 2007, Crowe et al., 1996, Morita et al., 1995), the orientation of the lower extremity (Haddad et al., 2006; Patterson et al., 2010; Andres and Stimmel, 1990, Shorter et al., 2008, Crenshaw and Richards, 2006) and EMG activities of the muscles (Isakov et al., 2000) while the use of angular rate of the lower extremity to identify gait asymmetry has not been reported. Therefore, this paper proposes the application of wireless miniature gyroscope as new approach to measure the lower limbs motion in walking. It also introduces Normalized Cross-correlations (Cc_norm) and Normalized Symmetry Index (SI_norm) to identify gait asymmetry and to overcome the limitations posed by the conventional methods.