Caracterización y clasificación de señales eléctricas cerebrales para aplicaciones BCI (Interfaz Cerebro Computador).
In this end-degree proyect, different feature extraction and pattern classification techniques were explored for the processing of alpha waves and sensorimotor rhythms in order to evaluate the posibility of using these methods in BCI applications. The brain signals used were acquired through EEG in...
Saved in:
| Hovedforfatter: | |
|---|---|
| Format: | bachelorThesis |
| Sprog: | spa |
| Udgivet: |
2017
|
| Fag: | |
| Online adgang: | http://dspace.unl.edu.ec/jspui/handle/123456789/19487 |
| Tags: |
Tilføj Tag
Ingen Tags, Vær først til at tagge denne postø!
|
Tilføj Tag | Summary: | In this end-degree proyect, different feature extraction and pattern classification techniques were explored for the processing of alpha waves and sensorimotor rhythms in order to evaluate the posibility of using these methods in BCI applications. The brain signals used were acquired through EEG in both cases. The work started with a state of the art review of BCI technology for identifying the most recommended techniques, used in the analisis of the chosen signals. The features used for alpha rhythms identification were stadistic parameters computed from the signal’s wavelet transform and a neural network was used as classifier. This method achieved high levels of accuracy when detecting alpha activity. The test waves used were from a healthy adult woman who gave her consent prior to the study. With these techniques a simple real-time BCI was developed. The goal of the system is to activate a serie of cues when the subject is capable of generating alpha waves. The sensorimotor rhythms analisis was based in the discrimination between left hand motor imagery and right hand motor imagery. Three feature extraction methods were evaluated for this purpose. Each of these methods had diferent results but finally one was found with enough accuracy for its possible BCI implementation. The features extracted of each method included those used for the alpha waves detector, AR coefficients and spatial filtering based in Common Spatial Patterns. A neural network was used for classification. The signals used for the sensorimotor rhythms processing were taken from the IIIA and 2A datasets of the BCI competitons III and IV, respectively. The signals of 12 test subjects were analized in total. |
|---|