Toward improving the visual stimulus meaning for increasing the P300 detection

TL;DR

This paper introduces a novel P300 speller paradigm (XP300) with visual feedback to enhance stimulus meaning, resulting in higher detection accuracy and user convenience compared to the classical approach.

Contribution

The main novelty is adding visual feedback to each stimulus in the P300 speller, improving detection performance and user experience.

Findings

XP300 achieves an average recognition rate of 85.25% versus 77.25% for CP300.

Single-trial detection is significantly better with XP300, with a mean AUC of 0.86.

XP300 is more convenient and user-friendly, enabling longer sessions.

Abstract

The P300 speller is a well known Brain-Computer Interface paradigm that has been used for over two decades. A new P300 speller paradigm (XP300) is proposed. It includes several characteristics: (i) the items are not intensified by using rows and columns, (ii) the order of the visual stimuli is pseudo-random, (iii) a visual feedback is added on each item to increase the stimulus meaning, which is the main novelty. XP300 has been tested on ten healthy subjects on copy spelling mode, with only eight sensors. It has been compared with the classical P300 paradigm (CP300). With five repetitions, the average recognition rate across subjects is 85.25% for XP300 and 77.25% for CP300. Single-trial detection is significantly higher with XP300 by comparing the AUC (Area Under Curve) of the ROC (Receiver Operating Characteristic) curve. The mean AUC is 0.86 for XP300, 0.80 for CP300. More importantly, XP300 has also been judged as more convenient and user-friendly than CP300, hence being able to allow longer sessions.