Maryam Misba, Afshan Fatima, Ruqiya Fatima 03 08 For prompt management and better patient outcomes early and precise diagnosis of Alzheimers disease AD especially the shift from Cognitively Normal CN to Mild Cognitive Impairment MCI is essential This study creates an improved model based on the Xception architecture building on previous deep learning techniques that use convolutional neural networks CNNs with channel attention mechanisms Both local and global neuroimaging features are captured by the suggested system using sophisticated feature extraction techniques which are then combined via a learning fusion process With a binary classification accuracy of 99 for CN versus MCI individuals our model performs noticeably better than earlier approaches This enhancement highlights the Xception networks capacity to identify subtle imaging indicators linked to early cognitive deterioration The system seeks to offer a dependable accessible diagnostic tool to assist medical professionals in the early detection of Alzheimers disease enabling prompt and focused treatment approaches Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026