The Classification of Alzheimer's Disease Using Artificial Intelligence

Chew, Jordan (2022) The Classification of Alzheimer's Disease Using Artificial Intelligence. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text JORDANCHEW_ Thesis.pdf Restricted to Registered users only Download (8MB)

Abstract

Image classification of medical image plays an important role in diagnostics for the early stage of the Alzheimer disease. Although the method to cure Alzheimer disease still undiscovered, but the early diagnosis benefit to the people to have early treatment to increase the survival rate. The objective of this project is to design and construct a Convolutional Neural Networks to classify the stage of demented level of Alzheimer disease on the Magnetic Resonance Imaging (MRI) images. The stage contains 4 classes of Alzheimer disease including Mild Demented, Moderate Demented, Nondemented and Very Mild Demented. The problem statement of this project is the diagnostic capability of brain scans in getting an error result for training using Convolutional neural networks model due to the complexity of the brain. For example, there are several brain diseases such as Mental Disorders, stroke and brain cancer, Convolutional neural networks model may detect the abnormal of the brain, but it cannot differentiate the type of brain disease. The major issue from the existing models is the MRI image contain large number of the feature representations. Besides that, some of the MRI Scan sample images still trained with limited amount of training set sample that will affect the expected result accuracy. Apart from that, the accuracy of the classification results was still one of the common challenges during constructing the Convolutional neural networks model. Therefore, the construction Convolutional Neural Networks using transfer learning approach had been introduced in this project. The performance of the Convolutional Network model has achieved a 99.72% accuracy to perform the Alzheimer disease classification. Next, the implementation of the selected Convolutional Neural Networks had been deployed into the web application had made the contribution of this project present in a more systematic way and more compatible for the user in future. In conclusion, the contribution of this project is important in the medical field that required a reliable technique that help the diagnose life-threatening diseases.