MRI IMAGE CLASSIFICATION MODEL FOR DETECTION OF ALZHEIMER’S DISEASE USING TRANSFER LEARNING

This study focuses on developing an effective Alzheimer's disease (AD) classification model using MRI images and transfer learning. Targeting individuals aged 65 and above, affected by the predominant form of dementia, the research employed an Alzheimer's Disease MRI Image dataset from Kaggle. Model selection involved options like EfficientNetB1, B3, B5, B7, VGG16, and VGG19, considering transfer learning or modifications. Two scenarios with distinct batch sizes (10 and 20) were explored in the model creation process. Evaluation, using a confusion matrix, determined that the EfficientNetB5 model yielded the highest accuracy at 99.22%, surpassing other models such as EfficientNetB1, B3, B7, VGG16, and VGG19. Notably, this research highlights the superior performance of EfficientNet over VGGNet in transfer learning for analyzing Alzheimer's disease MRI images. The study concludes with the implementation of a simple web system for testing model outcomes. Overall, the investigation underscores the efficacy of Convolutional Neural Network (CNN) modeling in Alzheimer's disease analysis and identifies EfficientNetB5 as the optimal model for accurate classification.