EEG analysis of gait movement preparation in the normal state and in the abnormal state of Parkinsonʼs disease with freezing of gait /
- Author(s): Velu, Priya D.;
- et al.
The cortical control of gait is an important aspect of locomotive function in healthy and diseased states. Here we used electroencephalography (EEG), signal processing, and machine learning methods to capture neural signals related to movement preparation of gait in healthy controls and in Parkinson's disease patients with freezing of gait (FOG). We focused on pre-movement EEG in tasks that required natural ambulation of the subjects through the environment with the ultimate goal of application to brain computer interfaces. First we aimed to predict the intent to start gait before the onset of any movement in normal, healthy individuals, and to distinguish this signal from different actions such as standing and pointing. Wavelets were used as features in LDA classification, which resulted in errors as low as 17% when averaged across nine subjects. The neural signal used for classification mostly consisted of contributions from slow cortical potentials (0.1-1 Hz) with additional, smaller contributions from mu (8-13 Hz) and beta (14-25 Hz) frequency bands from channels located over sensorimotor cortex. We then examined this pre-movement period in subjects who experience freezing of gait (FOG) to determine how PD FOG patients differ from healthy controls in baseline neural activity and how these measures change with visual modulation of the environment. The spectral activity and connectivity in and between occipital, parietal, and motor regions were assessed by using EEG signals recorded at channels Oz, Pz, and Cz in a network analysis of causal information flow. The results suggest possible pathological over-binding of cortical areas important for sensory integration and motor control through increased baseline theta and beta oscillations in freezing of gait patients who respond to visual feedback