چكيده به لاتين
P300 Speller as a most commonly used brain–computer interface (BCI) has been able to provide simple communication capabilities for people with severe motor or speech disabilities in order to have a better interaction with the outer world over the past years. Checker-board paradigm (CBP) is one of the most practical alternatives for row-column paradigm, enhancing the performance of the speller by preventing row-column induced errors. In this study, we developed a new paradigm to enhance the performance of the checker-board paradigm by alter presenting of an emoji stimulus instead of flashing the characters in checker-board paradigm. Also, in an offline analysis after finishing signal acquisition, by using the discrete wavelet transform (DWT) in feature extraction phase, the wavelet coefficients were used as the classifier input instead of using time samples. The performance of the proposed paradigm was evaluated and compared to the checker-board paradigm in an online experiment using FLD classifier over ten healthy subjects. For each paradigm, the recorded data from an offline session was used to calibrate the speller classifier; and consequently, the classification accuracy was calculated over online sessions. The proposed paradigm, showed 14% enhancement in classification accuracy with respect to the checker-board paradigm. The results of this study obviously showed that the stimuli obtained by presenting emoji instead of character flashing, effectively improved the speller classification accuracy After choosing the best mother wavelet from a group of common wavelet functions that created the best classification accuracy on the training data, the classification accuracy enhancement of 5.21% and 7.14% compared to using time samples were resulted for checker-board paradigm and checker-board with emoji stimulus paradigm, respectively. Also the results show that simultaneous use of checker-board with emoji stimulus and wavelet features, has resulted classification accuracy enhancement up to 20.44% with respect to classical method of checker-board paradigm with the time domain samples which is a very significant and remarkable difference. Totally in the current study, accuracy and speed of P300 speller is increased significantly using a new stimulus paradigm and wavelet features in feature extraction.
Keywords: Brain-Computer interface (BCI), P300 Speller, Checker-Board Paradigm, Emoji, Discrete Wavelet Transform(DWT)
