DISSERTATIONS

Data Assimilation for Quantitative Electroencephalogram Analysis.

Ph.D. in Intelligent Systems.

Advisors: Gibran Etcheverry and Rocio Salazar-Varas

Available here.

ABSTRACT

Human beings have a particular interest in knowing themselves and their connection with the environment in which they live. An example of this search of knowledge will be the study of the brain. Even though we have created tools and approaches to study its structure and complexity, we still have doubts regarding its performance in daily tasks, emotions, and feelings. To understand these phenomena, neuroscience, along with computational methods, evaluates brain images and electrical activity. One of the techniques used for brain signal analysis is data assimilation (DA), which combines numerical modeling and observations that best represent the state of a system. Although DA has been used to analyze the brain, other disciplines such as aerospace, navigation systems, and meteorology have had more development and study with this technique.

In this thesis, we present a methodology that combines the use of DA along with quantitative electroencephalogram (QEEG) techniques. For this, we selected as DA method the ensemble Kalman filter (EnKF) and the coherence and power spectrum as the QEEG analysis approaches. The former is used for the selection of the most relevant sensors, while the latter aims to evaluate a significant change in the frequency bands. With this information, we can start to understand how sensors from different brain areas contribute or not with the corresponding spectrum.

The methodology introduced on this thesis is meant to work as a supporting tool for the analysis of cognitive ability, motor ability, neurological disorders, among others. To demonstrate the viability of the proposed methodology, we used two datasets of electroencephalography (EEG) signals that dealt with the concentration and the learning of a new skill. The obtained outcomes show the advantages of using EnKF along with the QEEG methods for the study of EEG phenomena.

Malvasia: Brain-Computer Interfaces for enhancing museum’s accessibility.

B.Eng. in Systems and Information Technology.

Advisor: J. Alfredo Sánchez

Available here.

ABSTRACT

This thesis presents Malvasia, a system proposal where the users can interact with the pieces exhibited in a museum using only their brain signals through a Brain-Computer Interface (BCI). It is important to remark that any user could use this system, regardless they have a motor disability or not.

To explain its functionality, we present storyboards, a low-fidelity prototype, the design of the interfaces, and their architecture, including administrators for each component. Based on Malvasia, a prototype called M1 was created, which consists of the visualization and manipulation of a 3D model with the Emotiv EPOC +.

Besides, this prototype is used for user experience tests to evaluate the following points: Type of training, Program response, Scenarios to use the BCI, Impossible scenarios to use the BCI, Possible scenarios for the BCI in the museum, and Acceptance of the BCI. Thanks to these tests, we can affirm that BCI can be considered an interaction tool for these institutions.