Scholarlink

<p>Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and memory loss. Early and accurate detection of AD is crucial for effective treatment and intervention. This research communication paper proposes a novel approach for detecting Alzheimer’s disease using brain magnetic resonance imaging (MRI) images and a deep learning model built by integrating skip connections with the DEMentia NETwork (DEMNET) model. The proposed model leverages the power of deep neural networks to automatically learn discriminative features from brain MRI images and classify them as among 4 classes—Non Demented, Mild Demented, Very Mild Demented and Moderate Demented. Skip connections (SC) are incorporated into the network architecture to enable efficient information flow and alleviate the vanishing gradient problem, thereby enhancing the model’s performance in capturing both local and global image features. The model was trained on ADNI Dataset and Kaggle Dataset with 4 classes. Each class was balanced such that each class has 3200 images.</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

SC-DEMNET: Skip Connections Integrated DEMNET Framework for Alzheimer’s Disease Detection from Brain MRI Images

  • 期刊论文

  • 发表日期：

    2025年10月27日

查看全文
  • Manas Ranjan Prusty,
  • Hritik Goel,
  • Rishik Kumar

摘要

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline and memory loss. Early and accurate detection of AD is crucial for effective treatment and intervention. This research communication paper proposes a novel approach for detecting Alzheimer’s disease using brain magnetic resonance imaging (MRI) images and a deep learning model built by integrating skip connections with the DEMentia NETwork (DEMNET) model. The proposed model leverages the power of deep neural networks to automatically learn discriminative features from brain MRI images and classify them as among 4 classes—Non Demented, Mild Demented, Very Mild Demented and Moderate Demented. Skip connections (SC) are incorporated into the network architecture to enable efficient information flow and alleviate the vanishing gradient problem, thereby enhancing the model’s performance in capturing both local and global image features. The model was trained on ADNI Dataset and Kaggle Dataset with 4 classes. Each class was balanced such that each class has 3200 images.