Decoding p300 evoked potentials for brain computer interfaces (bci) aimed at assisting potential end-users at home

  1. Corralejo Palacios, Rebeca
Dirigida por:
  1. Roberto Hornero Sánchez Director

Universidad de defensa: Universidad de Valladolid

Fecha de defensa: 19 de enero de 2016

Tribunal:
  1. Laura María Roa Romero Presidente/a
  2. María García Gadañón Secretaria
  3. Jesús Poza Crespo Vocal
  4. Nuria Mendoza Laíz Vocal
  5. María Dolores del Castillo Sobrino Vocal

Tipo: Tesis

Resumen

A Brain Computer Interface (BCI) is a communication system that provides an alternative channel to natural communication and control processes since it bypasses the body’s normal efferent pathways, which are the neuromuscular output channels. Hence, a BCI system monitors brain activity and translates specific signal features, which reflect the user’s intent, into commands that operate a device. BCIs are especially interesting for severely impaired individuals that cannot perform physical movements. Conventional augmentative and alternative communication technologies require some form of muscle control and thus may not be useful for those with the most severe motor disabilities, such as late-stage amyotrophic lateral sclerosis (ALS), brainstem stroke or severe cerebral palsy. This Thesis proposes the use of P300 evoked potentials as signal control in BCI systems for end-users, i.e. severely impaired people. Selective attention to a specific and infrequent flashing symbol evokes a brain pattern called P300, which appears around the central and parietal brain cortex, about 300 ms after the stimulus presentation. A novel assistive BCI tool for environment control at home is researched. The proposed application is based on P300 evoked responses to infrequent stimuli, known as the ‘oddball’ paradigm. P300-based BCIs could be the most proper type of BCIs for severely disabled populations since it does not require training period. Furthermore, its typical paradigm allows to select in a fast manner the desired symbol among a large number of choices only by focusing on it. The methodology proposed in this work is focused on the real end-users. Thus, the design, the experiments and the assessment are centred on the potential end-users’ needs. A population of 30 potential BCI end-users, people with severe impairments due to different pathologies and degree of both motor and cognitive disabilities, took part in the study. Participants were recruited through the Spanish National Reference Centre on Disability and Dependence (León, Spain). The main needs identified for the population under study were related to three main categories: comfort, communication and security. Hence, the proposed assistive BCI tool was designed to cover them. Specifically, it enables to manage 8 devices, usually present at home, by means of 115 infrared control commands and 22 navigation commands. In order to simplify the interface, these commands were distributed among ten matrices: a main menu and nine sub-menus. In the experimental phase, the individuals interacted with the assistive tool during three sessions managing real electronic devices. Twenty three out of the thirty subjects were able to suitably manage the proposed BCI tool with accuracy higher than 80%. Nineteen out of them even achieved accuracy above 95%. Moreover, maximum rates up to 6.8 item/min and 25.9 bit/min were reached. These results show that most of potential end-users would be able to manage typical P300-based BCIs without special considerations. Furthermore, the analysis of the results suggested that the degree of impairment would not be a relevant issue in order to properly manage a P300-based BCI tool, since it seemed not to have an influence on the performance of individuals managing the proposed assistive tool. Statistical significance was not reached (p>0.05) between the online BCI performance and the degree of motor impairment, cognitive impairment and sustained attention ability of the thirty participants. Therefore, these applications could be appropriate for assisting and increasing the personal autonomy of severely impaired subjects at home. Furthermore, the proposed tool could easily increase and adapt the range of needs that can be satisfied for each specific patient. Thus, the dependence from caregivers, relatives and nurses could be reduced for disabled populations. The end-users’ assessment was carried out by means of a questionnaire. It showed that BCI potential users are willing to take part in this kind of studies, providing their requirements and their point of view. Hence, future works should focus on end-users for designing, developing and assessing assistive BCI applications. The population under study showed its satisfaction with the proposed assistive BCI tool through the assessment questionnaires. Most of the participants rated favourably the design and usefulness of the assistive tool for daily use at home. More specifically, users stated that the proposed tool was easy to manage and understand. Furthermore, participants considered that the proposed tool was really useful to increase their autonomy at home. They were even interested in including more home devices such as blinds, doors or bells; whereas they were not especially interested in BCIs for mobility applications. Nevertheless, individuals stated that electroencephalogram (EEG)-based BCIs are not practical on a daily basis. Participants suggested the use of wireless caps and dry electrodes in order to make EEG recordings more comfortable and practical for its regular use at home. Hence, future BCI developments for disabled populations should take these suggestions from real BCI end-users into consideration. In this Thesis, a novel P300-based assistive BCI tool for severely impaired people was investigated. Not only the design but also the experimental procedure and the evaluation were carried out through real end-users. Moreover, the experimental stage was performed with real devices and in the patients’ usual environment. Therefore, this study takes a noteworthy step considering the opinion and feelings of end-users as well as their degree of satisfaction.