Transfer Learning with Convolutional Neural Networks: a method for the medical diagnosis of Alzheimer's disease stage categorization
Main Article Content
Abstract
Alzheimer's disease (AD) is a neurological illness that progresses irreversibly and causes memory loss quickly. An accurate diagnosis is necessary for the proper management and treatment of AD, a devastating neurological condition. For Alzheimer's disease to be effectively treated, a prompt diagnosis is essential. Early diagnosis of Alzheimer's disease is crucial for the development of successful therapies and, ultimately, for providing optimal patient care. To diagnose distinct phases of Alzheimer's patients, such as CN, MCI, and AD, using MRI images, this article offers a thorough and up-to-date description of deep models (DL). Along with their techniques for choosing the dataset, pre-processing, and data analysis, the DL models that the researchers have determined are tested. Using the AD benchmark dataset, we assessed our technique's performance. More accurately than state-of-the-art techniques, the recommended approach identifies Alzheimer's disease. In this study, we pre-trained deep models using transfer learning to categorize Alzheimer's disease MRI data into several stages. The diagnosis of AD has been the subject of numerous transfer learning-based research efforts. A transfer learning-fused Inception-v3 with CNN model for classification was presented in this work. According to the results of our experiments, the suggested model outperforms existing models in the field in terms of prediction accuracy.The work concludes by outlining opportunities and pathways for more research, including developing specialized architectures, exploring fresh CNN modalities and applications, and applying transfer learning to AD imaging. The study underscores the immense potential of CNN and the significance of transfer learning-fused Inception-V3 in AD MRI imaging. However, it also acknowledges the necessity of further research and development to overcome existing hindrances and limitations. Correct AD classification has important therapeutic ramifications, including early detection with 97.10% accuracy results.