<![CDATA[Early identification of Mild Cognitive Impairment (MCI) is essential for preventing or delaying the progression of severe neurodegenerative disorders. The primary objective of this study is to develop an automated and computationally efficient framework for detecting MCI using structural brain imaging. The proposed research focuses on improving early diagnostic capability through a deep learning–based classification system that analyzes structural changes in brain images. The major contribution of this work lies in combining region-focused morphometric analysis with lightweight convolutional neural network architecture to achieve accurate classification while maintaining computational efficiency suitable for clinical environments. The methodology involves extracting anatomically meaningful features from structural brain scans using a region-of-interest based morphometric approach. Brain images undergo several preprocessing procedures including skull stripping, normalization, spatial alignment and data augmentation to ensure consistency and robustness of the dataset. After preprocessing, the images are used to train a lightweight deep learning model that performs binary classification between cognitively normal subjects and individuals with MCI. The study employs a publicly available neuroimaging dataset consisting of structural brain scans and associated clinical information. Experimental results demonstrate that the proposed framework achieves strong classification performance while maintaining low computational complexity. The model achieves 88.2% subject-wise test accuracy and 0.90 cross-validation accuracy, outperforming commonly used architectures such as VGG16 (78.1%) and ResNet50 (53.7%). These findings indicate that lightweight neural networks combined with region-based anatomical analysis can effectively support automated screening of MCI. The proposed approach has potential implications for scalable clinical decision support systems and may assist neurologists in early diagnosis, timely intervention, and improved cognitive healthcare management. Future research may explore multimodal data integration and longitudinal clinical validation to further enhance diagnostic reliability.]]>
