Technight27/ADHD_detection

The objective of this project is to propose a deep learning oriented methodology for real-time identification of ADHD patients from their brain's EEG signals.

ADHD_detection

The objective of this project is to propose a deep learning oriented methodology for real-time identification of ADHD patients from their brain's EEG signals.The dataset comprised 60 healthy subjects and 61 ADHD patients’ EEG data. According to the 10-20 standard system, 19 electrodes were placed on the relevant part of the human scalp for the record of the EEG. To eliminate the relevant noise, artifacts and noise that originates from the subjects’ movements and blinking, band-pass filtering method is applied in the range of 0.1 Hz to 50 Hz. The filtered data of EEG was then employed for training the ADHD-Net for the automated identification of ADHD patients. For this purpose, 70% of the dataset was used for training, 10% for validation and the rest 20% was used for testing. The proposed model consists of CNN layers, BiLSTM layers, and an attention layer. The CNN layers help in the process of feature extraction from the samples of the EEG data. The BiLSTM layers extract yet more features from the features already extracted by CNN. The attention mechanism improves the vital attributes that have been learned by the BiLSTM layers. To this end, two fully dense layers are employed for classification, to help sort classified data out from the rest.