In the rapidly evolving digital age, human-machine interface technologies are continuously being improved. Traditional methods of computer interaction, such as a mouse and a keyboard, are being supplemented and even replaced by more intuitive methods, including eye-tracking technologies. Conventional eye-tracking methods utilize cameras to monitor the direction of gaze but have their limitations. An alternative and promising approach for eye-tracking involves the use of electroencephalography, a technique for measuring brain activity. Historically, EEG was primarily limited to laboratory conditions. However, mobile and accessible EEG devices are entering the market, offering a more versatile and effective means of recording bioelectric potentials. This paper introduces a gaze localization method using EEG obtained from a mobile EEG recorder in the form of a wearable headband (provided by BrainBit). The study aims to decode neural patterns associated with different gaze directions using advanced machine learning methods, particularly neural networks. Pattern recognition is performed using both ground truth data collected from wearable camera-based eye-tracking glasses and unlabeled data. The results obtained in this research demonstrate a relationship between eye movement and EEG, which can be described and recognized through a predictive model. This integration of mobile EEG technology with eye-tracking methods offers a portable and convenient solution that can be applied in various fields, including medical research and the development of more intuitive computer interfaces.
The review focuses on the most promising methods for classifying EEG signals for non-invasive BCIs and theoretical approaches for the successful classification of EEG patterns. The paper provides an overview of articles using Riemannian geometry, deep learning methods and various options for preprocessing and "clustering" EEG signals, for example, common-spatial pattern (CSP). Among other approaches, pre-processing of EEG signals using CSP is often used, both offline and online. The combination of CSP, linear discriminant analysis, support vector machine and neural network (BPNN) made it possible to achieve 91% accuracy for binary classification with exoskeleton control as a feedback. There is very little work on the use of Riemannian geometry online and the best accuracy achieved so far for a binary classification problem is 69.3% in the work. At the same time, in offline testing, the average percentage of correct classification in the considered articles for approaches with CSP – 77.5 ± 5.8%, deep learning networks – 81.7 ± 4.7%, Riemannian geometry – 90.2 ± 6.6%. Due to nonlinear transformations, Riemannian geometry-based approaches and complex deep neural networks provide higher accuracy and better extract of useful information from raw EEG recordings rather than linear CSP transformation. However, in real-time setup, not only accuracy is important, but also a minimum time delay. Therefore, approaches using the CSP transformation and Riemannian geometry with a time delay of less than 500 ms may be in the future advantage.
1 - 2 of 2 items