Artificial Intelligence in ADHD Diagnosis Using CNN

Modern medicine has a challenge in gathering, evaluating, and applying the vast amount of knowledge needed to address challenging clinical issues. The development of AI systems for medical applications has been linked to the development of medical AI. They are made to aid the physician in making a diagnosis, selecting a course of treatment, and anticipating results. Artificial Neural Networks (ANNs), fuzzy expert systems, evolutionary computation, and hybrid intelligent systems are some examples of such systems. By processing EEG signals through two-dimensional colored picture transforms with GASF, the study aims to give early identification for Attention Deficit Hyperactivity Disorder (ADHD), one of the most prevalent neurobehavioral diseases. The EEG DATA FOR ADHD / CONTROL CHILDREN dataset of kids with ADHD disorder provided the EEG data. ADHD is a hotly debated topic among medical professionals who contend that the condition is underdiagnosed and that many kids go undiagnosed. Therefore, it is crucial to bring such topic up and to find ways to help in easing the diagnosis procedure of this disease. In this master's thesis, the 2D colored image data was trained and tested using the AlexNet deep learning model. Alexnet is an eight-layer Convolutional Neural Network (CNN) model. The main objective is to motivate researchers in medical image interpretation to extensively rely on CNNs in their analysis and diagnosis. In the database, mentioned above, 16 channels were used. Among those 7 channels were used and those are Fp1, Fp2, F3, Fz, F4, P3, and P4. Selected channels' GASF pictures were utilized to train the AlexNet model. For training and testing, the generated model achieved accuracy of 99% and 71.7%, respectively.