Learning Discriminative Spatiospectral Features of ERPs for Accurate Brain–Computer Interfaces

Abstract

Constructing accurate predictive models is at the heart of brain-computer interfaces (BCIs) because these models can ultimately translate brain activities into communication and control commands. The majority of the previous work in BCI use spatial, temporal, or spatiotemporal features of event-related potentials (ERPs). In this study, we examined the discriminatory effect of their spatiospectral features to capture the most relevant set of neural activities from electroencephalographic recordings that represent users' mental intent. In this regard, we model ERP waveforms using a sum of sinusoids with unknown amplitudes, frequencies, and phases. The effect of this signal modeling step is to represent high-dimensional ERP waveforms in a substantially lower dimensionality space, which includes their dominant power spectral contents. We found that the most discriminative frequencies for accurate decoding of visual attention modulated ERPs lie in a spectral range less than 6.4 Hz. This was empirically verified by treating dominant frequency contents of ERP waveforms as feature vectors in the state-of-the-art machine learning techniques used herein. The constructed predictive models achieved remarkable performance, which for some subjects was as high as 94% as measured by the area under curve. Using these spectral contents, we further studied the discriminatory effect of each channel and proposed an efficient strategy to choose subject-specific subsets of channels that generally led to classifiers with comparable performance.

Design and Optimization of a BCI-Driven Telepresence Robot Through Programming by Demonstration

“…On average, all users were able to exert seven control commands per minute to interact with and control the telepresence robot. Although the primary goal of the present work is not enhancing BCI decoders, it is interesting to observe that the overall performance of the constructed BCI system is close to the state-of-the-art reported in the BCI literature [1], [35], [36], [55], [56].…”

“…• Spectral Filtering: EEG data has been band-pass filtered between 0.5-12 Hz range using a Fourier filter. First, the signal was Fourier transformed, and then a weighting is applied to suppress and remove unwanted frequencies outside the frequency of interest range [35]. The weighted signal was inverse Fourier transformed to obtain the filtered signal.…”

“…In [35], [39], we showed that Logistic Regression with L 2 -Ridge (LRR) penalty [40] [43, p. 190-191], and logistic regression with L 1 -least absolute shrinkage and selection operator penalty [44]. We observed this relative superiority of LRR on both extracted spatiospectral features and temporal features of ERP waveforms [35], [39]. In what follows, we describe the application of LRR on temporal features, which, as observed in [39], generally lead to a higher…”

Biomarkers and neuromodulation techniques in substance use disorders

“…Recent advances in the development of machine learning algorithms in the context of EEG-based brain machine interfaces used e.g. logistic regression, Bayes estimation, support vector machines (Abibullaev and Zollanvari 2019), convolutional or recurrent neural networks (Lawhern et al 2018;Roy et al 2019) to identify specific electrophysiological neural features in real-time using continuously recorded neural activity. Brain potentials thereby classified as "pathological" could be used to adjust stimulation to normalise neural activity and improve behaviour control enabling an individually and situationally adapted intervention (Campanella 2013).…”

Effects of Visual Attention on Tactile P300 BCI

“…Finally, these commands can be used for practical applications including, but not limited to, wheelchair navigation [4,5], character speller [6,7], and robotic arm manipulation [8,9]. e brain activities that are most frequently used to control BCI systems include event-related potentials (ERP) [10], steady-state evoked potentials [11], and event-related desynchronization (ERD) [12] and event-related synchronization (ERS) [13]. In an ERP-based BCI, the induction of the ERP is achieved by presenting a predictable sequence of stimuli with one or more rarely, randomly occurring (unpredictable) stimuli interleaved amongst the predictable stimuli.…”