A Brain–Computer Interface Speller with a Reduced Matrix: A Case Study in a Patient with Amyotrophic Lateral Sclerosis

Abstract

Visual P300-based Brain–Computer Interface (BCI) paradigms for spelling are aimed at offering a non-muscular communication channel for those people with severe motor impairment, such as locked-in patients. To be as effective as other assistive technologies, these systems have to achieve a greater communication rate. One way to do so is to develop better interfaces. In this regard, we thought of using a 4 x 3 symbol matrix based on the T9 interface developed for mobile phones. Due to presenting a reduced matrix and relying on an adaptation of the T9 predictive text system, we expected that this speller would provide a higher communication rate than usual 6 x 6 matrix spellers that are based on Farwell and Donchin’s classic proposal. As a proof of concept, a locked-in patient with amyotrophic lateral sclerosis tested our T9-like visual BCI speller along with two different 7 x 6 conventional matrix spellers. The comparison of her performance results with those of a sample of three healthy participants suggested that it was possible for this locked-in patient to control the T9-like speller as well as they did, and thus, write a target sentence considerably faster than when she used the alternative spellers