Learning EEG-based spectral-spatial patterns for attention level measurement

Abstract

Abstract-In our every day life, our brain is constantly processing information and paying attention, reacting accordingly, to all sorts of sensory inputs (auditory, visual, etc.). In some cases, there is a need to accurately measure a person's level of attention to monitor a sportsman performance, to detect Attention Deficit Hyperactivity Disorder (ADHD) in children, to evaluate the effectiveness of neuro-feedback treatment, etc.In this paper we propose a novel approach to extract, select and learn spectral-spatial patterns from electroencephalogram (EEG) recordings. Our approach improves over prior-art methods that was, typically, only concerned with power of specific EEG rhythms from few individual channels. In this new approach, spectral-spatial features from multichannel EEG are extracted by a two filtering stages: a filter-bank (FB) and common spatial patterns (CSP) filters. The most important features are selected by a Mutual Information (MI) based feature selection procedure and then classified using Fisher linear discriminant (FLD). The outcome is a measure of the attention level.An experimental study was conducted with 5 healthy young male subjects with their EEG recorded in various attention and non-attention conditions (opened eyes, closed eyes, reading, counting, relaxing, etc.). EEGs were used to train and evaluate the model using 4x4fold cross-validation procedure. Results indicate that the new proposed approach outperforms the prior-art methods and can achieve up to 89.4% classification accuracy rate (with an average improvement of up to 16%). We demonstrate its application with a two-players attention-based racing car computer game.

Inter-subject transfer learning with an end-to-end deep convolutional neural network for EEG-based BCI

“…Using two filters in a row including filter bank (FB) and common spatial pattern (CSP), they extracted spectral-spatial features from EEG which was recorded using multiple electrodes placed in various brain regions. Then, the extracted features were sent to a fisher linear discriminant (FLD) classifier for classification task [23]. Their approach outperformed the conventional methods based on only spectral features.…”

Using Recent BCI Literature to Deepen our Understanding of Clinical Neurofeedback: A Short Review

“…In the future, such tools might lead to improved individualized NF features, with a higher SNR, going beyond heuristics like Individualized Alpha Frequency (IAF) (Bazanova et al, 2010). Moreover, ML algorithms are also used to automatically identify functionallyspecific cortical sources, e.g., SMR sources (Blankertz et al, 2008, Vidaurre et al, 2012 or attention-related sources (Hamadicharef et al, 2009). Such sources are defined as the individual EEG channels (Lal et al, 2004;Arvaneh et al, 2011) or combination of channels, i.e., spatial filters (Blankertz et al, 2008;Lotte & Guan, 2011;Samek et al, 2014), which record the greatest modulation of EEG activity during a given task.…”

Spatio-Temporal Detection of Divided Attention in Reading Applications Using EEG and Eye Tracking

“…Berka et al developed one of the first commercial systems, known as B-Alert [6], to detect states of attention (high-engagement, low-engagement, relaxed wakefulness, and sleepy) in real time by a medical grade EEG device, but were not configured to detect divided attention. Hamadicharef et al were the first to apply the filter bank common spatial pattern (FBCSP) [8] algorithm to an attention task with EEG data. They found that the FBCSP method classified up to 89.4% between states of attention and relaxation using a 15-channel medical grade EEG device.…”