Influence of Gait Cycle Normalization on Principal Activations

Abstract

The Clustering for Identification of Muscle Activation Pattern (CIMAP) algorithm has been recently proposed to cope with the high intra-subject variability of muscle activation patterns and to allow the extraction of principal activations (PAs), defined as those muscle activation intervals that are strictly necessary to perform a specific task. To assess differences between different PAs, gait cycle normalization techniques are needed to handle between- and within-subject variability. The aim of this contribution is to assess the effect of two different time-normalization techniques (Linear Length Normalization and Piecewise Linear Length Normalization) on PA extraction, in terms of inter-subject similarity. Results demonstrated no statistically significant differences in the inter-subject similarity between the two tested approaches, revealing, on the average, inter-subject similarity values higher than 0.64. Moreover, a statistically significant difference in the inter-subject similarity among muscles was assessed, revealing a higher similarity of PAs extracted considering the distal lower limb muscles. In conclusion, our results demonstrated that PAs extracted from healthy subjects during a walking task at comfortable walking speed are not affected by the time-normalization approach implemented.