Decoding of motor information from noninvasive electroencephalographic signals for brain-computer interfaces

Abstract

Brain-Computer Interfaces (BCIs) are emerging assistive technologies that provide an artificial communication pathway between the brain and the external world. These systems translate a mental task performed by the user into commands to control external devices using brain signals recorded with invasive or noninvasive techniques. This is remarkably interesting for different applications related with neuromotor rehabilitation field, for example, BCI systems for neurorehabilitation therapies where BCIs provides patients with motor impairments with a non-muscular communication channel that could be used to activate a robot-assisted rehabilitation device. However, there are other applications not related to the neurorehabilitation field where this technology provides an enhancement for the communication between the user and the its environment. An example of this is the BCI’s for automotive applications, where BCI technology is applied as a part of Advanced Driving Assistance Systems to avoid crash vehicle situations. Irrespective of the type of application, movement-related BCI systems use the motor imagery (MI) paradigm as the mental task that the user performs and which the system detects and classifies by generating commands to drive external devices Despite the success of Motor Imagery-based BCI systems, there are some characteristics of these interfaces that are susceptible to be improved. First, to improve the performance of mental task detection, novel classification models can be explored to compare their performance with the conventional classification models used in BCI (such as Linear Discriminant Analysis (LDA) and Support Vector Machines (SVM)). Secondly, there are applications in which the motor imagery paradigm has limitations that avoid the BCI system to be able to detect multiple mental motor tasks related to diverse movements generated by the same limb. In addition, the MI paradigm is not fully adaptable to detect intentions to execute sudden movements, which is important for applications where the objective of BCI is to support and complement the rehabilitation therapies for people with the ability to recover their physical motor functions. Finally, the validation of neurorehabilitation therapies based on BCI online for end users (people with motor disabilities). It is necessary to evaluate the usefulness of this technology in the rehabilitation of patients with motor disabilities. This PhD thesis investigates the detection of information related to movements from non-invasive EEG signals exploring potential solutions to the limitations of conventional Motor BCI systems. The first study explores novel classification models as those based on deep learning which could improve the BCI system robustness and performance. This study aims to compare classical and Deep Neural Networks (DNN) algorithms for the recognition of Motor Imagery (MI) tasks from electroencephalographic (EEG) signals. The second study investigates the detection of emergency braking from driver’s electroencephalographic (EEG) signals that precede the brake pedal actuation. EEG signals were classified using support vector machines (SVM) and convolutional neural networks (CNN) in order to discriminate between braking intention and normal driving. The third study assess the feasibility of recognizing two rehabilitative right upper-limb movements from pre-movement EEG signals. These rehabilitative movements were performed self-selected and self-initiated by the users using a motor rehabilitation robotic device. We proposes diverse anticipatory detection scenarios that discriminate EEG signals corresponding to non-movement state and movement intentions of two same-limb movements. Finally, the last study is focused on the development of a BCI-driven functional electro-stimulation system (FES) aimed at neurorehabilitation of the upper limbs of patients with spinal cord injuries (SCI). Furthermore, clinical benefits of the BCI-FES system in SCI patients are explored by estimating quantitative EEG parameters for motor rehabilitation.