ADDNet - An Enhanced Convolutional Neural Network for Detection and Classification of Alzheimer's Disease

The World Health Organization (WHO) reports that Alzheimer's disease (AD) is the principal factor that causes dementia, diminishing cognitive abilities of people across the globe. Early detection of Alzheimer's disease (AD) using non-invasive methods is critical for long-term human health. Health and wellbeing for all is one of the sustainable development goals set by the United Nations (UN). In line with this goal, there has been a significant research effort in the healthcare domain to detect Alzheimer's disease. Methodologies based on learning for automatic diagnosis of Alzheimer's disease (AD) have increased in significance with the development of artificial intelligence (AI), machine learning (ML), and deep learning (DL). Existing AI-enabled methods for medical image processing modalities such as magnetic resonance imaging (MRI) are efficient because they are based on profound understanding techniques like convolutional neural networks (CNN). Motivated by this, our investigation on CNN-based methods revealed that there is a need for leveraging network efficiency with the configuration of layers and optimizations. At the end of this research, we provide a system that can automatically detect and classify AD. The framework is based on a deep knowledge of how to detect AD in a given patient using a magnetic resonance imaging (MRI) image. The proposed CNN model, known as Alzheimer's Disease Detection Net (ADDNet), enhances the baseline CNN model. This model has improved the architecture for progressive generation of characteristics and enhanced it for early AD detection. The suggested method, called learning-based Alzheimer's disease detection (LbADD), makes use of ADDNet. We conducted our empirical research using the frequently used Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset as a benchmark. The results of our experiments demonstrate that ADDNet clearly demonstrates its robustness, as it achieves an overall accuracy of 98.83%, which is superior to that of other models. The Clinical Decision Support System (CDSS) can incorporate our approach in healthcare units to assist doctors in AD detection, diagnosis, and correlation of facts.