Pseudo-online Multimodal Interface Based on Movement Prediction for Lower Limbs Rehabilitation

Abstract

Patients with physical disabilities can benefit from robotic rehabilitation since it can provide more control, accuracy and variety of training modes. This improves the efficiency of the patient’s rehabilitation. This work presents a multimodal platform for acquisition and processing of EEG and EMG signal with inertial sensors data in order to detect lower limb movement intentions. A pseudo-online technique was also developed. Experiments were conducted with 5 healthy subjects performing lower limb motor tasks and an experimental protocol is proposed. In the future, this interface will be integrated in an active knee orthosis for robotic rehabilitation. The results obtained show that the system is capable to acquire, process and classify the signals synchronously. The movement intention was evaluated taking into account EEG and EMG signal together with a OR logic, detecting $60.0\pm 21.2\,\%$ of movement intentions. The movement anticipation achieved $881.5\pm 136.3$ ms based on EEG signal and $137.6\pm 77.5$ ms based on EMG signals.