Tesis de Maestría / master Thesis
Evaluation of electroencephalogram source localization methods for the decoding of motor information
Fecha
2020-07Registro en:
Esparza Lepe, R. A. (2020). Evaluation of electroencephalogram source localization methods for the decoding of motor information (Tesis de maestria). Instituto Tecnológico de Estudios Superiores de Monterrey,
925384
Autor
Esparza Lepe, Roberto Alejandro
Institución
Resumen
There are several technologies in the world that start using brain signals as input, such as wheelchairs, gadjets, or prosthesis. This machines are known as brain-computer interfaces, which are systems that establish a means of communication between a user and his surroundings using only neural activity. Traditionally, the electroencephalogram (EEG), which is the effect produced by brain activity, is recorded through electrodes placed on the head. Features are extracted from the EEG in order to find patterns using a machine learning algorithm. And finally, commands that activate an application are generated from these patterns. However, BCIs have reached a limit in their performance, so it has been proposed to use the neural sources that generate the EEG to improve their reach. This thesis explores the use of EEG source localization methods (EEG-SLMs) to determine if they are a viable option for future research. Three studies were carried out in order to evaluate EEG-SLMs. The first study evaluates the performance of EEG-SLMs using synthetic signals in a simulated environment. The second study evaluates the performance of EEG-SLMs using real EEG signals obtained from an experiment with motor tasks. The third study evaluates the classification accuracy of machine learning algorithms when features are extracted from the EEG and when they are extracted from the neural sources that generated the EEG. From the results obtained, the performance of the EEG-SLMs was verified when tested with simulated scenarios and known synthetic signals in the first study. The second study was useful in learning about the main brain areas involved in the development of motor tasks, these being not only the primary and secondary motor cortices, but also the areas responsible for locating the body and objects in space, as well as those responsible for calculating distances and take decisions. Furthermore, satisfactory results were obtained in the classification of motor tasks using neural sources.