Using Artificial Neural Networks on Multi-channel EEG Data to Detect the Effect of Binaural Stimuli in Resting State

Abstract

More than 7% of the population of the world is afflicted by anxiety disorders. If related tomood disorders, anxiety can trigger or escalate other symptoms and affects mental diseases, akin depression, and suicidal behavior. Recent works have shown that binaural beats are able to reduce anxiety and modify other psychological conditions, significantly changing cognitive processes and mood states. They can be defined as a low-frequency acoustic stimuli perceived when a given individual is subjected to two marginally different wave frequencies, from 200 to 900 Hz. In the present work, we applied a 5 Hz binaural beat to 6 different subjects, to detect if relevant changes could be noticed in their brain waves after the stimuli. Twenty minutes stimuli in ten separate sessions were applied. In order to detect these possible differences, we used a single hidden layer Multi-Layer Perceptron neural network and compared its results to non-parametric statistical tests and Low-Resolution Brain Electromagnetic Tomography (eLORETA). Results obtained on eLORETA point to a strong increase in the current distribution, mostly in the Alpha 2 band, at the Anterior Cingulate, pertinent to the recognition and expression of emotions, as well as the monitoring of mistakes regarding social conduct. Many Artificial neural networks models, principally Multi-Layer Perceptron architectures, proved to be able to highlight the main differences with high separability in Delta and Theta spectral bands.