Characterization of gait maturation using frequency domain analysis of CoM acceleration

Introduction: The use of inertial sensors, placed at approximately CoM level (L5 level), is today a common practice in gait analysis [1]: CoM acceleration data allow indeed evaluating spatio-temporal gait parameters [2] and characterizing different populations [3]. Frequency domain analysis of the CoM acceleration signal has been already proposed in literature, in order to assess the range of frequency spectrum and its dependency on the motor task [4–6] or in order to evaluate characteristics of the signal in relation to task event (e.g. impact-related shock, Parkinson's freezing) [7,8]. The aim of this study was to investigate the use of the frequency domain analysis of CoM acceleration in characterizing gait maturation from childhood to adulthood, evidencing deterioration of gait performance during ageing. Methods: The study was conducted on: 7 toddlers at two weeks of walking experience (T2wks, 13 ± 2 mo, 10 ± 2 kg, 78 ± 4 cm), 7 toddlers at 6 months of walking experience (T6mo, 18 ± 2 mo, 11 ± 2 kg, 82 ± 2 cm), 7 4-year old children (4YC, 4 y, 16 ± 2 kg, 101 ± 3 cm), 7 6-year old children (6YC, 6 y, 23 ± 1 kg, 121 ± 2 cm), 7 adolescents (15YA, 15 y, 60 ± 13 kg, 162 ± 6 cm), 7 young adults (25YA, 25 ± 1 y, 171 ± 9 cm, 67 ± 14 kg) and 7 elderlies (E, 75 ± 7 y, 77 ± 11 kg, 167 ± 6 cm). The participants performed an instrumented over ground gait task wearing a tri-axial wireless inertial sensors (OPALS, Apdm, USA) (fs 128 Hz) located on L5. Ten consecutive strides, for all the participants, were analyzed [9]. The FFT of the acceleration signals was used to determine the frequency corresponding to the maximum of the monolateral signal spectrum (fMAX). The signal power (PW) was calculated integrating the potential spectrum density (psd), obtained with the Matlab built-in function pwelch. Then PW was normalized with the maximum value of it, obtaining the normalized signal power (PW%). The frequencies corresponding to the 50% (f50%) and 98% (f98%) of the PW%. were calculated. fMAX, f50% and f98% were calculated along the three axes (AP, ML V). The estimated parameters were tested with gaussianity test in such way was possible performed the right statistical analysis: Kruskal–Wallis test with a p-value of 5%. When age effect was found, a multiple comparison test was performed. Results: Kruskal–Wallis results showed effect of age on all the features in V direction and on f98% and f50% in AP direction. No significant results were found in ML direction. In Fig. 1a and b are showed 25° 50° and 75° percentiles of fMAX_AP and fMAX_V respectively. Colours indicate differences found by the multiple comparison test for T2wks group (blue): red indicates statistical differences and grey no statistical differences. Fig. 1 Download high-res image (67KB)Download full-size image Fig. 1. Discussion: The results of the present work indicate that f50%, f98% and fMAX can be relevant and descriptive parameters of gait development. In particular, fMAX_AP, changes from high to low values with the increasing of the age and could describe improvement and deterioration of the motor development; moreover fMAX_V and f98%_V discern between immature and mature gait during gait development and could be used as an index of deterioration of gait performance during ageing.