From the classical analysis of the center of pressure signals to a novel approach: study of rotational components.

The balance control is known to be a complex motor skill, which involves the integration of many types of sensory information. The postural control is achieved by feedback mechanisms based on the body-sway motion detected primarily by visual, vestibular, and proprioceptive sensory systems. Static posturography provides an objective assessment of postural control by characterizing the body sway during upright standing. The Center-of-Pressure (CoP) signal is recorded by a force platform and analyzed according to different techniques. Traditional approaches decompose the CoP signal into antero-posterior (AP) and medio-lateral (ML) time series and extrapolate geometrical, statistical and spectral parameters. In this dissertation the problem of analyzing CoP signals was faced with two different strategies. Firstly, I applied geometrical parameters to describe the CoP signals and a multivariate statistical approach to analyze differences in quiet standing among groups of subjects, in two different studies. A new acquisition protocol was proposed, which adds to frontal open- and closed-eye conditions, conditions in which quiet standing of the subject is evaluated after a fast or a slow head rotation, on the left or on the right side, both with eyes open and closed. The aim of the first work was to analyze the postural control of volleyball players and the impact that vision has on it. The main hypothesis of this study was that, since volleyball players use the visual system differently from untrained subjects, the role of vision on postural control should also be different. Volleyball players showed greater CoP ellipses with respect to controls. A multivariate approach to data analysis demonstrated that the two groups were different when the subjects kept their eyes open, but they were not with visual deprivation. The influence of the athlete’s expertise and team role on balance performances was also analyzed. Differences in the upright stance of national and regional athletes were found, as well as differences between defensive players and hitters. The second work evaluated differences in postural performances in controls and patients with residual neuro-ophthalmic deficits after a traumatic brain injury. It was possible to evidence significant balance abnormalities in TBI patients with respect to controls. Moreover, by means of a multivariate analysis, I was able to discriminate different levels of residual neuro-ophthalmic impairment. The second approach was based on the hypothesis that the CoP signal contains rotational components. To verify this hypothesis and to extract rotational components from the CoP signal I applied the rotary spectra analysis, a well known technique developed in the meteorological and oceanographic field. Rotary spectra analysis involves the representation of a two-dimensional signal in the complex plane as a superimposition of ellipses, which can be analyzed in terms of their shape and orientation. Each ellipse is the sum of a counterclockwise (CCW) and a clockwise (CW) rotating phasors, called rotary components. This approach allows to consider both the AP and ML time-series not only as mono-variate signals, but also as components of a complex signal, taking into account both the amplitude of the whole signal and its phase. The rotary spectra approach permits to separate rotational isofrequential components from non-rotational ones, providing a different approach in understanding of the physiological mechanisms underlying postural control. The rotary spectra analysis was applied to the CoP signal of a population of healthy subjects. The presence of rotational components in the signal was demonstrated, and useful parameters about the rotational characteristics of the body sway were extracted. An interesting result highlighted by this new approach was that the mode value of the rotary spectra fell in the range 0.14-0.17 rps. The peak that was observed in this frequency range probably has a physiologically explainable meaning that was never documented before. I hypothesized that rotary spectra peaks obtained in the study of postural control during upright standing are strictly correlated to the bursts of muscle sympathetic nerve activity. This work passed the first stage of review to be published in “Motor Control”. Since the CoP signal is not strictly stationary, considering the classical rotary spectra analysis it is possible only to obtain frequency marginals of the signal during the 60s test. To analyze the CoP trajectories as non-stationary random signals, the classical rotary spectra theory was extended, to deal with non-stationary signals. In particular, I applied both bilinear Cohen's class Choi-Williams time-frequency distribution and wavelet method to obtain time-frequency analysis of rotating components. This allowed evaluating rotational characteristics of the CoP signal in the time-frequency plane. As interpretation and classification of a large number of time-frequency distributions could be difficult, time expensive, and highly depending on the researcher, I developed an automatic analysis method. It consists of three main steps: (i) using image processing tools, the main frequency components were identified on the time-frequency plane; (ii) a set of features was extracted; (iii) a clustering algorithm was applied. Applying this method to the distributions calculated by the time-frequency rotary analysis of the CoP signals of a population of healthy subjects, I obtained a fast and repeatable description of each distribution in terms of frequency components. Moreover, a classification in five groups of the distribution was obtained. In conclusion, both followed strategies gave satisfactory results. The proposed protocol and the evaluation of geometrical parameters allowed highlighting differences between groups of pathological and healthy subjects. I proposed a new approach in analyzing CoP signal in terms of rotational components, which gave interesting results on a group of healthy subjects. The technique was extended to time-frequency domain and an automatic analysis method of time-frequency distribution was developed. The rotary spectra analysis, both in stationary and non-stationary cases, is complementary to other approaches described in literature to study the CoP signal, and could contribute in reaching better understanding of the physiological mechanisms underlying postural control.