Motor imagery BCI system with visual feedback: Design and preliminary evaluation

Abstract

Nowadays, strokes are a growing cause of mortality and many people remain with motor sequelae and troubles in the daily activities. To treat these sequelae, alternative rehabilitation techniques are needed. This article describes the design, development and preliminary evaluation of a system based on Brain Computer Interfaces (BCI) by Motor Imagery, with visual feedback for lower limb rehabilitation of people post stroke. The system consists of three modules: Sensing and Conditioning; Control Signal Generator; and Visual Feedback. The first module acquires, filters and segments 5 channels of EEG. The second module performs spatial filtering using a Laplacian, estimates the signal power spectral density, extracts and selects EEG features which are then used by the classifier to detect event related desynchronization. The command signal generated by the BCI is inputted into the third module, which simulates the movement of foot dorsiflexion of an avatar displayed on a screen. For the implementation, the BCI2000, V-REP platforms and MATLAB software were used. Performance evaluation of the system was done in a healthy volunteer by estimating the sensitivity and specificity, and through interviews with specialists. Average values for sensitivity and specificity were 0,67 and 0,70 respectively, and professional opinions were very good. These results are encouraging for deepening the performance evaluation system and taking steps for clinical implementation.