An alternative approach to normalization and evaluation for gait patterns: Procrustes analysis applied to the...

In order to compare gait patterns, a common procedure is to normalize strides both in time and magnitude. The stride duration is usually normalized to a time percentage before averaging curves. As the timing of event occurrences may shift across strides, the shape of the averaged curves is distorted and therefore the standard deviation is overvalued. Stride magnitude normalization is performed by means of dimensionless numbers. However, there is little agreement on which body size correction methods should be used. The Procrustes method describes curve shape and shape change in a mathematical and statistical framework, independently of time and size factors. The present study aims to explore how this technique may be used for time- and magnitude-stride normalization to reflect individual and group mean responses. The Procrustes method, which combines quantitative and visual features, is applied to the shape of the ankle and knee cyclograms. Superimposition of 25 cyclograms (10 for sprinters (SP) and 15 for middle-distance runners (MDR)) was supplemented by statistical procedures (principal component analysis, discriminant function) to extract the main key events, which vary according to the athletic specialities. In comparison with the MDR (poulaine-shaped cyclogram), the ovoid cyclogram of SP reveals the following gait indicators: a short braking phase, a rapid initial lower limb swing in the forward direction, a fast upward movement of the knee and ankle, and an active foot contact. The Procrustes approach could be used to describe other quasi-periodic movements through relative motion plots (e.g., cyclograms, angle-angle diagrams, phase plane portraits).