The Incorporation of Artificial Intelligence in the Identification of Neurological Disorders

Abstract

With the rise of neurological disorders among a wide age bracket today, efficient and accurate diagnosis has faced some challenges. Due to the amount of time needed to observe the physiological symptoms of the brain as well as the behavior of the patient, certain neurological disorders can be mistaken for another. This led to the initial research question: To what extent can AI limit the amount of misdiagnoses by improving efficiency and reducing diagnosis time for neurologists in the United States? This study focused on monitoring three different artificial intelligence models and their efficiency to provide an initial diagnosis of a disorder based on an MRI scan. The three models that were compared were the Convolutional Neural Network (CNN), Multilayer Perceptron (MLP), and Recurrent Neural Network (RNN). Due to the similarity between the functioning of the CNN model and the typical human brain, it was hypothesized that the CNN model would be the most efficient. After running the brain scan through each model multiple times and averaging the data, it was found that the Convolutional Neural Network had the quickest response time and the most accuracy compared to the other models. This response time was based on the trained AI model’s ability to make a diagnosis. A multiclass output was utilized for the final diagnosis results. Further implementation of AI models in the diagnosis process, especially CNN models, can lead to significant improvement in the field of neuroscience.