BD2EMNET: An Innovative Strategy for Managing Multi-class Classification and Class Imbalance Issues in Alzheimer’s Disease

Alzheimer’s Disease (AD) stands as a progressive neurodegenerative ailment that significantly affects the well-being of those afflicted and their families. The prompt and precise identification of the condition plays a pivotal role in effectively managing and intervening in the disease progression. Nonetheless, the task of categorizing AD across various stages encounters complexities due to uneven distributions of data, where the less common categories frequently suffer from insufficient representation. In this paper, we introduced an innovative architecture BD2EMNET (DEMentia NETwork including Borderline SMOTE and DenseNet-121 architecture) tailored to address the issue of imbalanced classes and multi-class classification issues, thereby enhancing the accuracy in the classification of a five-class AD dataset. Our proposed architecture underwent evaluation using a comprehensive five-class ADNI dataset. Comparative experiments clearly showcased the superior performance of our approach when compared to cutting-edge models, resulting in a substantial accuracy improvement. The novel architecture put forth achieved an impressive overall accuracy rate of 99.88%, showcasing its exceptional proficiency in identifying the diverse stages of AD with remarkable accuracy. The significance of our research goes beyond the surface, as it establishes a sturdy foundation for precise and early-stage classification of AD, holding the potential to greatly benefit early intervention strategies and patient care practices. In conclusion, the paper introduces an inventive architecture that adeptly addresses the challenge of imbalanced class distribution and multi-class classification tasks, leading to a substantial enhancement in the accuracy of AD classification within a five-class dataset. This contribution bears the potential to advance the early detection of the disease, thereby fostering improved outcomes for patients in the battle against AD.