A Hybrid Deep Learning Framework to Predict Alzheimer’s Disease Progression Using Generative Adversarial Networks and Deep Convolutional Neural Networks

• Published in: Arabian journal for science and engineering (2011) Vol. 49; no. 3; pp. 3267 - 3284
• Main Authors: SinhaRoy, Rajarshi, Sen, Anupam
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.03.2024; Springer Nature B.V
• Subjects: Artificial neural networks; Deep learning; Engineering; Generative adversarial networks; Humanities and Social Sciences; Machine learning; Magnetic resonance imaging; Medical imaging; multidisciplinary; Neural networks; Research Article-Computer Engineering and Computer Science; Science; Healthcare; Generative adversarial networks; Alzheimer’s disease; Convolutional neural networks; Super-resolution GANs; Deep convolutional GANs
• ISSN: 2193-567X; 1319-8025; 2191-4281
• DOI: 10.1007/s13369-023-07973-9

A major research subject in recent times is Alzheimer’s disease (AD) due to the growth and considerable societal impacts on health. So, the detection of AD is essential for medication care. Early detection of AD is critical for effective treatment, and monitoring the time period between normal aging’s unavoidable cognitive loss and dementia’s more catastrophic degradation is common practice. The deep learning method for early diagnosis and automated categorization of AD has suddenly gained a lot of attention since rapid advancement in the field of GANs approaches has now been used in the clinical research sector. Many recent studies using brain MRI images and convolutional neural networks (CNNs) to identify Alzheimer’s disease have yielded promising results. Instead of adequately engaging with the lack of real data, many research papers have focused on prediction. The main purpose of this paper is to do this by generating synthetic MRI images using a series of DCGANs. This paper demonstrates the effectiveness of this concept by cascading DCGANs that imitate different stages of Alzheimer’s disease and utilizing SRGANs to enhance the resolution of MRI scans. The purpose of this research is to come forward and tell if an individual might just get Alzheimer’s disease. CNN, DCGANs, and SRGANs are used in this paper to present a deep learning-based approach that improves classification and prediction accuracy to 99.7% and also handles the lack of data and the resolution of data.