Estimation of attention deficit and hyperactivity disorder (ADHD) with artificial neural networks using EEG signals

ESTIMATION OF ATTENTION DEFICIT AND HYPERACTIVITY DISORDER (ADHD) WITH ARTIFICIAL NEURAL NETWORKS USING EEG Mustafa BAŞARAN Advanced technologies, Master Thesis, 2019 Thesis Supervisor: Assist. Prof. Dr. Mustafa TOSUN SUMMARY Attention Deficit Hyperactivity Disorder (ADHD) is a neurological disorder characterized by hyperactivity, carelessness and sudden behavior. EEG signals are also frequently used for disease diagnosis. Bioelectrical signals that occur as a result of neural activity of the brain are called electroencephalogram (EEG) signals. EEG signals are not periodic. Phase, amplitude and frequencies change continuously. EEG frequency bands are called delta, theta, alpha and beta. In individuals with attention deficit and hyperactivity disorder (ADHD), the power densities in the frequency bands vary compared to normal individuals. In this study, the power spectrum intensities of EEG signals obtained from patients with attention deficit and hyperactivity disorder were obtained by using the welch method. The power spectrum density values were applied to the Forward Feed Back Propagation (FFBPNN) Artificial Neural Network, Elman Network and Self Organizing Maps (SOM) network as feature values of EEG signals. Trained networks were tested with test data. As a result of the test, the success of classification in FFBPNN network was 89%, 84% in Elman network and 70% in SOM network. Keywords: ADHD, Artificial Neural Networks, Electroencephalogram, SOM, Welch method.