Command control wi̇th brain computer interface (BCI)

The electroencephalogram (EEG) is a monitoring method that records the electrical activity of the brain, initially emerging in the field of Neuroscience to diagnose epilepsy disorders. In recent studies, EEG signals have been used in many investigations to detect an individual's response to specific stimuli, as they establish a direct pathway between the human brain and computers. In Brain-Computer Interfaces (BCI), electrical potentials generated in the brain serve as input signals. One of these signals is the steady-state visually evoked potentials (SSVEP). SSVEP-based BCI typically tracks brain activity by directing the user's attention to visual stimuli that flicker or change at specific frequencies. These visual stimuli are usually presented on a computer screen. The user's attention to these visual stimuli at specific frequencies in the brain causes changes in SSVEP signals. By detecting these SSVEP signals, a BCI system can predict the user's attention and provide an interface to execute specific commands. The study involved 8 participants with an average age of 25.62. Prepared visual stimuli consisted of right, left, forward, and backward arrow symbols presented to the user at frequencies of 7, 8, 9, and 10 Hz, respectively, with EEG recordings taken. EEG recordings were obtained using an Emotiv EPOC+ 14-channel headset with a sampling frequency of 256 Hz. The total EEG recording duration was 1152 seconds, with a recording duration of 144 seconds per person. Subsequently, the signals from channels P7, P8, O1, and O2 were filtered using a Butterworth bandpass filter in the 2-15 Hz range and normalized, then segmented into 5-second epochs for use in the study.Next, the signals were transformed into the frequency domain using the Fast Fourier Transform (FFT) method, and their Power Spectral Densities (PSDs) were obtained. In the classification phase, SVM (Support Vector Machines), ANN (Artificial Neural Networks), KNN (K-Nearest Neighbors), and NB (Naive Bayes) algorithms were applied for both binary and four-class classifications. An accuracy rate of 63.90% was achieved with the Decision Tree (DT) algorithm in the binary classification study. In the four-class classification study, an accuracy rate of 31.20% was reached using the Support Vector Machine (SVM) algorithm.