Nonlinear analysis of brain activity, associated with motor action and motor imaginary in untrained subjects

Identification of brain activity associated with motor execution and, more importantly, with motor imagery is necessary for the development of brain-computer interfaces. Most of recent studies were performed with trained participants which demonstrated that the motor-related brain activity can be detected from the analysis of multichannel electroencephalograms (EEG). For untrained subjects, this task is less studied, but at the same time much more challenging. This task can be solved using the methods of nonlinear dynamics, allowing to extract specific features of the neuronal network of the brain (e.g., the degree of complexity of EEG signals and degree of interaction between different brain areas). In this work, we analyze the spatio-temporal and time-frequency characteristics of the electrical brain activity, associated with both the motor execution and imagery in a group of untrained subjects, by applying different methods of nonlinear dynamics. At the first stage, we apply multifractal formalism to the analysis of EEG signals to reveal the brain areas which demonstrate the most significant distinctions between real motor actions and imaginary movement. Then, using time-frequency wavelet-based analysis of the EEG activity, we analyze in detail the structure of considered brain areas. As a result, we distinguish characteristic oscillatory patterns which occur in different areas of brain and interact with each other when the motor execution (or imagination) takes place. Finally, we create an algorithm allowing online detection of the observed patterns and experimentally verify its efficiency.