Scholarlink

This study aims toward providing a comprehensive exploration of AI's capabilities in classifying Alzheimer's disease using brain MRI images. The study meticulously examines a wide spectrum of deep learning models, encompassing custom-designed CNN architectures, transfer learning models (VGG16, VGG19, ResNet, MobileNetV2, InceptionV3, DenseNet169, EfficientNetb0), and traditional machine learning models (Random Forest, SVM, Logistic Regression). Besides that, the study forays into the development of hybrid models, in the hope of synergistically combining the strengths of deep learning and other robust algorithms such as XGBoost, Gaussian Naive Bayes, and SVM. Training and testing of models are conducted on a dataset of 6400 preprocessed MRI images, which are properly labeled into four different classes: Mild Demented, Moderate Demented, Non-Demented, and Very Mild Demented. In order to ensure a proper and fair comparison of the models, the study employs a set of performance metrics, such as accuracy, precision, recall, and AUC. This research highlights the immense potential of AI in revolutionizing the early screening and diagnosis of Alzheimer's disease. By providing accurate and timely diagnostic support, the AI models meticulously crafted in this study have the capacity to significantly enhance the capabilities of healthcare professionals, enabling them to make well-informed decisions and devise effective treatment strategies.

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

A Novel Approach for Neurodegenerative Disorder Detection Using Custom CNN Architecture

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Davinder Paul Singh,
  • Dev Patel,
  • Debabrata Swain,
  • Pawandeep Kour,
  • Shubham Mahajan,
  • Amit Kant Pandit

摘要

This study aims toward providing a comprehensive exploration of AI's capabilities in classifying Alzheimer's disease using brain MRI images. The study meticulously examines a wide spectrum of deep learning models, encompassing custom-designed CNN architectures, transfer learning models (VGG16, VGG19, ResNet, MobileNetV2, InceptionV3, DenseNet169, EfficientNetb0), and traditional machine learning models (Random Forest, SVM, Logistic Regression). Besides that, the study forays into the development of hybrid models, in the hope of synergistically combining the strengths of deep learning and other robust algorithms such as XGBoost, Gaussian Naive Bayes, and SVM. Training and testing of models are conducted on a dataset of 6400 preprocessed MRI images, which are properly labeled into four different classes: Mild Demented, Moderate Demented, Non-Demented, and Very Mild Demented. In order to ensure a proper and fair comparison of the models, the study employs a set of performance metrics, such as accuracy, precision, recall, and AUC. This research highlights the immense potential of AI in revolutionizing the early screening and diagnosis of Alzheimer's disease. By providing accurate and timely diagnostic support, the AI models meticulously crafted in this study have the capacity to significantly enhance the capabilities of healthcare professionals, enabling them to make well-informed decisions and devise effective treatment strategies.