A Comprehensive Review of Convolutional Neural Network Applications for Alzheimer’s Disease Diagnosis Using Neuroimaging

Alzheimer’s Disease (AD) is a progressive neurodegenerative disorder where early diagnosis is critical for effective management but remains a significant clinical challenge. The diagnosis often relies on identifying subtle morphological changes in the brain, a task for which Deep Learning (DL) models like Convolutional Neural Networks (CNNs) are exceptionally well-suited. This paper provides a comprehensive review of recent seminal studies that apply CNN architectures to AD classification using neuroimaging data. We systematically analyze a range of methodologies, from foundational 2D and 3D models to advanced hybrid frameworks incorporating attention mechanisms. The analysis confirms that advanced CNNs achieve high diagnostic accuracy, demonstrating their potential as powerful diagnostic tools. However, the field faces critical hurdles: the clinical translation of these models is hindered by the high computational cost of leading architectures, concerns over generalizability due to an over-reliance on a few public datasets, and a widespread lack of open-source code that severely limits scientific reproducibility and collaborative progress.