Title: Selection of effective probes for an individual to identify P300 signal generated from P300 BCI speller
Authors: Weilun Wang; Goutam Chakraborty
Addresses: School of Software and Information, Iwate Prefectural University, Iwate, Japan ' School of Software and Information, Iwate Prefectural University, Iwate, Japan
Abstract: P300 is a strong event related potential (ERP) generated in the brain and observed on the scalp when an unusual event happens. To decipher P300 signal, we have to use the property of P300 to distinguish P300 signal from non-P300 signal. In this work, we used data collected from P300 BCI Speller with 128 probes. Conventional BCI speller uses eight probes at pre-defined locations on the skull. Though P300 is strong in the parietal region of the brain, location of the strongest signal varies from person to person. The idea is that, if we optimise probe locations for an individual, we could reduce the number of probes required. In fact, the process mode for the raw brain wave signals also will affect the classification accuracy. We designed an algorithm to analyse the raw signals. We achieved over 81% classification accuracy on average with only three probes from only one target stimulus and one non-target stimulus.
Keywords: event related potential; ERP; probes reduction; P300 amplitude; brain computer interface; BCI.
International Journal of High Performance Computing and Networking, 2019 Vol.15 No.1/2, pp.72 - 79
Received: 30 Mar 2018
Accepted: 09 Nov 2018
Published online: 08 Nov 2019 *