Alzheimers Disease Identification using Transfer Learning and Advanced Convolutional Neural Network Models

Alzheimer's disease (AD) is an important global health issue that primarily affects older people. As the neurodegenerative disease advances, only two of the physical and mental issues that can develop are memory loss and language loss. The early and precise identification of AD has drawn increasing attention in recent years. This work suggests a deep learning-based approach for categorizing pictures of both healthy and Alzheimer's-affected brains. Image categorization and ROI extraction are the two main components of the suggested approach. Another method for diagnosing AD or moderate cognitive impairment (MCI) is ROI extraction, which focuses on separating global brain areas, such the hippocampus, from the whole brain in an imaging or scan visualization. Later on, classification may be done using transfer learning and convolutional neural networks (CNN). Important details are removed before the CNN model recognizes the images. For classification, Transfer Learning makes use of pre-trained AlexNet features. Tests using the OASIS (Open Access Series of Imaging Studies) dataset show that the Transfer Learning method outperforms the CNN-based method in terms of accuracy (92.86%).