Use this URL to cite or link to this record in EThOS:
Title: Advancing the P300 based BCI design
Author: Cota, Navin Gupta
Awarding Body: University of Essex
Current Institution: University of Essex
Date of Award: 2011
Availability of Full Text:
Access from EThOS:
Brain Computer Interface (BCI) systems capture brain signals and convert them into commands without using peripheral nerves or muscles. This doctoral thesis embarks on advancing the design of P300 based BCI system using electroencephalogram (EEG) responses to visual and audio stimuli. This research work investigates techniques for pre processing, feature extraction as well as attempts some variations in paradigm design from a P300/0ddball paradigm perspective. Pre-processing is an important module in BCI systems. This thesis proposes a fully automated method (GALME-=ICA) to reduce artefacts effectively for an offline scenario. The method uses the recorded EEG channels only and does not require recorded EOG channels. The P300/0ddball systems were designed so as to reduce perceptual errors, thereby making them participant friendly. Few variations in paradigm design . reported here have not been attempted before. In one study, the spatial effect of the target stimuli location with respect to the non-target stimuli was explored. In another study limiting interference from irrelevant task related stimuli was studied. Results are discussed in terms of classification accuracies and bit rates. This thesis also proposes the usage of gamma band features with P300 time domain features for RSVP and audio based oddball paradigms for BCI based applications. An attractive integration of EEG-NIRS for monitoring participant concentration was also attempted. Preliminary results highlight the importance in selecting training datasets for good online classification results from an EEG viewpoint. There is no denying the fact that recent advances in BCI field have led researchers to explore new applications like cursor control and game control. However, there are many challenging problems which remain to be solved, before a commercial BCI system becomes a reality. Hence there is an urgent need to explore better system design and develop novel signal processing as well as machine learning algorithms. Imagination combined with practicality appears to be the key here. This Doctoral work is a step forward in this direction.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available