Alzheimer's disease (AD) is a common chronic disorder with a high incidence rate that disproportionately affects elderly people. Deep learning (DL) has been increasingly popular in recent years, resulting in notable developments and innovations in medical imaging. Consequently, deep learning has emerged as the preferred approach for analyzing medical visuals, particularly in the realm of Alzheimer's disease detection. In this chapter, the authors performed a comparative analysis of various DL models such as convolutional neural networks, DenseNet, ResNet, EfficientNet, etc., which have shown some groundbreaking results for AD disease detection. Also, they focused on investigating data collection and feature extraction techniques pertinent to AD. In addition, they further discussed briefly the deep learning models for AD detection. This not only increases hope for the advancement of AD research and therapy but also highlights how deep learning can revolutionize the field of medical image analysis and illness identification.
Top1. Introduction
The prevalent form of dementia, known as Alzheimer's disease (AD), poses a significant healthcare challenge in the 21st century. AD stands as the sixth major contributor to death in the United States, impacting approximately 5.5 million individuals aged 65 and above. The cost of managing AD in the United States in 2018 was $277 billion, which included medical costs, social welfare costs, and wage losses for patient families (Alzheimer’s Association, 2018). Unfortunately, there are presently no recognized disease-modifying therapies for AD, an irreversible and progressive brain condition marked by a deterioration in cognitive ability (Destrooper & Karran, 2016). To delay or stop the course of the disease, much effort has been put into finding methods for early identification, especially during the pre-symptomatic phases.From a medical standpoint, disease is seen as an extra health problem rather than a fully scientific phenomenon. Chronic illnesses are a major ethnographic topic since each person has their own recollections and distinctive knowledge of them (Armstrong & Hilton, 2014). According to medical science, a disease is defined as a disruption of the body's normal function that manifests certain signs and symptoms (Armstrong & Hilton, 2014; Scully, 2004). To properly diagnose and treat the condition, clinicians must have a thorough understanding of the illness's mix of outward indications and symptoms. A complicated and difficult procedure, disease diagnosis in healthcare frequently involves several components and circumstances for medical symptoms (Leaman, 2013). Medical judgment must be formed by assessing the disease while taking into account numerous subjective and impartial elements. Since a treatment strategy cannot be decided upon until the diagnosis is confirmed, a prompt and correct diagnosis is crucial in the case of serious illnesses or disorders (Scheuermann, 2009).
To acquire correct findings and aid consultants in controlling illnesses in their early stages, current medical research initiatives employ artificial intelligence, most notably the deep learning model (Croft et al., 2015; Malmir, 2017; Nashi, 2018; Nilashi, 2017). Researchers from all around the world are investigating different neural network applications to forecast catastrophic illnesses. Since data is sometimes elusive and demands creative technology for exact analysis, academics are working on various formats to address the issue of evaluating disorders. To limit or stop the course of illnesses, much effort has been put into creating tools for early identification, particularly in pre-symptomatic phases (Galvin, 2017; Schelke, 2018). Advancements in advanced neuroimaging methods such as positron emission tomography (PET) and magnetic resonance imaging (MRI) have enabled the identification of structural and molecular biomarkers associated with Alzheimer's disease (AD) (Veitch et al., 2019). The fast advancement of neuroimaging has made it difficult to integrate massive amounts of high-dimensional, multimodal neuroimaging data. As a result, using computer-aided machine learning techniques for integrative analysis is becoming more and more popular. Popular pattern analysis techniques, including logistic regression, support vector machines, linear program boosting approaches, linear discriminant analysis, and support vector machine-recursive feature elimination (SVM-RFE), have shown promise. To effectively implement machine learning algorithms, it is crucial to establish appropriate architectural design and pre-processing protocols (Lu & Weng, 2007).