Integrated Feature-Based Predictive Classification of Alzheimer's Disease Using Transfer Learning, LBP, and Support Vector Machine

Alzheimer's disease (AD) is a neurodegenerative disorder affecting memory and cognition. Accurate diagnosis enables better disease management. This paper presents a hybrid approach for binary classification using structural magnetic resonance imaging (sMRI) scans from the OASIS dataset. ResNet-18 serves as a feature extractor, enhanced by Local Binary Patterns (LBP), with classification done via Support Vector Machine (SVM). The model achieved 95% accuracy in distinguishing Demented from Non-Demented subjects, supporting its potential for clinical application.