Abstract:
The field of brain-computer interfaces (BCIs) remains a hot research topic as these systems are not yet ready for daily use by amyotrophic lateral sclerosis (ALS) patients. The biggest challenge when employing BCIs for communication is a low information transfer rate (ITR) caused by the low number of classes and the error rate.
This dissertation examines solutions with healthy subjects and ALS patients, but the main focus of this work is on the patients with their varying degrees of disability. A magnetoencephalogram (MEG) study using a larger number of classes including cognitive tasks reached an offline improvement of 28% (in 10 healthy subjects) over the two-class BCI.
Automatic recognition of the error-related potential (ErrP), which is present when a user commits an error, lead to an online ITR increase of 31% (in 17 healthy subjects) and 21% (in 6 patients) in an electroencephalogram (EEG) study.
For the first time, two completely locked-in (CLIS) patients with chronically implanted electrocorticogram (ECoG) electrodes participated in long-term BCI and cognition detection tests. The breathrough in BCI communication with CLIS patients was not possible, although cognition in the form of significant oddball responses was detected in some of the sessions in one patient.