ARCHIVES

In the global growing rate of Alzheimer’s Disease (AD) has lead to transition from manual clinical diagnosis to high-precision, trustworthy automated neuroimaging diagnostics. It has been studied a variety of deep learning models and multi-modal graph convolutional networks, it has considerable issues in the research majorly class imbalance mainly for the lack of availability of Moderate Dementia samples especially for elderly data and the fundamental black-box attributes of neural network decision-making. To address this limitations, this paper shows that a Clinical Decision Making Support which converts the system passive to actionable by integrating with the Grad-CAM visual explainability is required. This analysis also indicated the use of Wasserstein Generative Adversarial Networks (WGAN) to improve the quality of the generated synthetics images of MRI to balance the dataset to fill the gap between algorithmic prediction and real-world clinical treatment.