Loading...

Design and Implementation of a P300 Speller System by Using Auditory and Visual Paradigm

Jalilpour, Shayan | 2019

488 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52376 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hajipour Sardouie, Sepideh
  7. Abstract:
  8. The use of brain signals in controlling devices and communication with the external environment has been very much considered recently. The Brain-Computer Interface (BCI) systems enable people to easily handle most of their daily physical activity using the brain signal, without any need for movement. One of the most common BCI systems is P300 speller. In this type of BCI systems, the user can spell words without the need for typing with hands. In these systems, the electrical potential of the user's brain signals is distorted by visual, auditory, or tactile stimuli from his/her normal state. An essential principle in these systems is to exploit appropriate feature extraction methods which can identify the P300 component with high accuracy. So, using a suitable feature extraction method has a vital role in the performance of BCI systems.Visual protocols can operate faster and more accurately than other protocols, which has led to most of the BCI research using visual stimulation. In cases that patients have poor vision, and they may not be able to use visual systems, hearing protocols are suggested which can be visual independent. In this project, two auditory and visual stimulation paradigms are presented for spelling letters. The auditory paradigm is such that a list of 30 stimulus words is presented to the user, and he/she must focus on words which contain the target letter. By testing the proposed protocol on one participant, the accuracy and information transfer rate (ITR) of 88% and 51.3 bits/minute are achieved, respectively, for 12 iterations of the target character.Also, the hybrid paradigm is proposed as the visual protocol, which combines the RSVP and SSVEP paradigms. By using RSVP stimulation, the target stimulus is detected, and the target character is identified by employing SSVEP stimulation. To evaluate the proposed protocol, the experiment was conducted on six participants and average accuracy and ITR of 93.66% and 41.33 bits/minute are obtained, respectively, by using five iterations of the stimuli. Finally, a new feature extraction algorithm based on the common spatial pattern (CSP) is proposed, so that it can accurately detect the character in the P300 speller systems. By using the regularization term in the process of solving the objective function and using spatial, temporal and frequency information of the data, the best result (91.7%) is obtained in comparison with other CSP based methods
  9. Keywords:
  10. P300 Speller ; Brain-Computer Interface (BCI) ; Event Related Potential (ERP) ; Steady State Visual Evoked Potential (SSVEP) ; Common Spatical Patterns

 Digital Object List

 Bookmark

No TOC