Scholarlink

<p>Alzheimer’s disease (AD) is increasingly prevalent, posing substantial challenges for families and healthcare systems worldwide. Automated diagnostic methods are essential, as manual AD diagnosis is labour-intensive and error-prone. This paper introduces a novel deep learning-based approach using convolutional neural networks (CNNs) to analyze brain MRI data for AD diagnosis. We utilize a large dataset comprising 6400 images across four AD stages: non-demented, very mildly demented, mildly demented, and moderately demented. We used a fully convolutional network (FCN) structure that enables the use of several encoder networks to extract features and classify them. Vigorous experimentation demonstrates that our model outperforms the state-of-the-art architectures, such as ResNet50 and ResNet101, and obtains an impressive 99.98% accuracy. This article shows the possibility of deep learning to revolutionize AD diagnosis by offering a powerful tool for identifying an early disease and its stage. The study is a breakthrough in solving the issue of AD diagnostics, providing improved treatment of patients and effective utilization of the healthcare system.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Advancing clinical Alzheimer’s diagnosis with end-to-end deep learning on MRI data

  • 期刊论文

  • 发表日期：

    2026年1月13日

查看全文
  • Harshvardhan Khimsuriya,
  • Mohd. Aquib Ansari,
  • Naveen Kumar

摘要

Alzheimer’s disease (AD) is increasingly prevalent, posing substantial challenges for families and healthcare systems worldwide. Automated diagnostic methods are essential, as manual AD diagnosis is labour-intensive and error-prone. This paper introduces a novel deep learning-based approach using convolutional neural networks (CNNs) to analyze brain MRI data for AD diagnosis. We utilize a large dataset comprising 6400 images across four AD stages: non-demented, very mildly demented, mildly demented, and moderately demented. We used a fully convolutional network (FCN) structure that enables the use of several encoder networks to extract features and classify them. Vigorous experimentation demonstrates that our model outperforms the state-of-the-art architectures, such as ResNet50 and ResNet101, and obtains an impressive 99.98% accuracy. This article shows the possibility of deep learning to revolutionize AD diagnosis by offering a powerful tool for identifying an early disease and its stage. The study is a breakthrough in solving the issue of AD diagnostics, providing improved treatment of patients and effective utilization of the healthcare system.