Archive | 2019
Preliminary Design of a Regions-Based BCI Speller for Locked-in Patients
Abstract
An important application for Brain-Computer Interfaces (BCI) is using them for spelling task in subjects with severe motor disabilities. Most BCI spellers are based on the matrix proposed by Farwell and Donchin [1] or variations of these. Several defects of those, were resolved by other authors. In this project, an improvement of the regions-based matrix of Fazel-Rezai [2] is proposed to better realize the user experience and the speller performance. This is a preliminary design of a region-based speller using BCI to record electroencephalographic (EEG) signals, taking in mind Locked-in patients. We developed a computer based system which generate a two levels stimulation matrix. The first level has ten (10) regions with alphabet characters, numbers, some short words and a region with the capacity of propose words or phrases, in base of previous letter or words spelled by the user. The second level has only six symbols or words/phrases belonging to the region selected by the patient in the first stimulation level. The Application Module of the system, designed to detect P300 signal, was developed using the BCI2000 software [3] and the framework BCPy2000 [4]. The results of preliminary in vivo trials showed that the enhanced matrix can elicit P300 and the recognition module exhibited an average accuracy in characters’ identification, ranging 75–89% and reaching 100% in one of the trials.