Alzheimer’s Diseases Detection by Using Xception Deep Neural Network with Data Augmentation

In our modern age, amidst the whirlwind of progress and change, there exists a silent epidemic that transcends time and technology, affecting individuals and families in profound ways: Alzheimer’s disease. This illness, which slowly causes memory loss and difficulty thinking, reminds us how delicate our minds are and how complex brain health can be. This work focuses on leveraging deep learning algorithms with data augmentation to enhance the accuracy of AD detection. The study examines four cutting-edge deep learning models: Xception, InceptionV3, EfficientNetB3, and ResNet50. Robust feature extraction from MRI images is accomplished by using ResNet50, which is renowned for its deep design with skip connections. EfficientNetB3, with its efficient design and high accuracy, contributes to effective classification of AD stages. InceptionV3’s multi-level feature extraction capabilities are utilized to capture intricate patterns indicative of AD progression. Xception’s depth wise separable convolutions enhance pattern recognition, aiding in precise disease detection. Utilizing deep learning algorithms, notably Xception, and employing data augmentation techniques, our study attained an exceptional accuracy of 97.18% in detecting Alzheimer’s Disease, utilizing the "Alzheimer’s Diseases Detection" dataset. In order to improve early diagnosis and treatment planning for Alzheimer’s patients, the research intends to give a thorough examination of these deep learning methods in AD detection.