Early Alzheimer's Detection: Comparing ResNet50 and EfficientNetB0 on MRI Data

Alzheimer's disease is a chronic neurodegenerative illness that leads to dementia and irreversible cognitive decline. Early detection is crucial for managing the disease's progression and improving the quality of life for affected individuals. In this study, a Convolutional Neural Network (CNN) is developed using 80,000 MRI brain images from the OASIS database to classify Alzheimer's patients into four severity groups: non-demented, very mild, mild, and moderate. The dataset exhibits significant class imbalance, which is addressed by downsampling the majority classes. The CNN models ResNet50 and EfficientNetB0 are employed for classification, achieving accuracies of 80.86% and 79.33%, respectively. Evaluation metrics such as precision, recall, and F1 score are calculated using confusion matrices, demonstrating the models' effectiveness in Alzheimer's disease classification. The study underscores the importance of early diagnosis and proposes deep learning techniques as valuable tools in medical imaging analysis for neurodegenerative disorders.