کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Applying Parameter-Oriented Learning to Identify Statistical EEG Features Associated with Depression

نویسندگان :

Sara Bargi Barkouk¹ Melika Changizi² Mahdi Zolfagharzadeh Kermani³ Ali Asadi Zeidabadi⁴

1- دانشگاه ازاد اسلامی واحد علوم و تحقیقات 2- دانشگاه ازاد اسلامی واحد علوم و تحقیقات 3- دانشگاه ازاد اسلامی واحد علوم و تحقیقات 4- دانشگاه ازاد اسلامی واحد علوم و تحقیقات

کلمات کلیدی :

computational parameters،depression،detection system،electroencephalogram،statistical features

چکیده :

Major Depressive Disorder (MDD) is a prevalent mental health condition with a complex neurophysiological basis. Identifying unique patterns of brain activity associated with MDD, with a special focus on electroencephalography (EEG), enables the ability to identify mental disorders by increasing the understanding of functional brain mechanisms. The present study introduces an innovative approach to design a depression detection system based on parameter-oriented learning. This approach results in developing a detection system using statistical features extracted from EEG signals. For this purpose, the processing unit of the system was first defined as a combination of calculation parameters, including long window duration, short window duration, and the temporal overlap percentage between periods, and then the measurement of statistical features extracted from the EEG signal was optimized based on these parameters. The best results were calculated using the five optimum features, which were related to four channels, along with the use of the K-Nearest Neighbors (K-NN) classifier, which resulted in obtaining the accuracy, F1-score, and area under curve (AUC) as 95.85%, 95.40%, and 95.78%, respectively.