Scholarlink

<p>The use of intelligent diagnostic methods for early Alzheimer’s disease detection carries significant implications across various domains, encompassing economic impact and treatment opportunities. This study focuses on Alzheimer’s disease diagnosis and disease progression investigation utilizing transfer learning. Our proposed method employs a transfer learning approach following the acquisition of MRI images from the ADNI database, leveraging the advantage of low computational cost. For preprocessing, the proposed method employs FreeSurfer software, employing a slice-based processing approach along the Axial, Coronal, and Sagittal axes. Three pre-trained models are utilized for feature extraction. Subsequently, an ensemble learning technique combines the strengths of these models across all axes simultaneously. This approach effectively addresses two critical issues in this domain. Upon classifying the three stages of AD, MCI, and CN, the results demonstrate a reasonable diagnostic performance for Alzheimer’s disease. Particularly noteworthy is the classification accuracy of 99.84% along the axial axis in examining disease progression between pMCI and sMCI stages. The presented algorithm facilitates the estimation of disease progression two years prior to Alzheimer’s disease onset without additional computational burden, demonstrating highly competitive accuracy compared to other existing methods.</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

Early Diagnosis from Mild Cognitive Impairment to Alzheimer’s Disease Dementia on MRI Images Using Transfer Learning

  • 期刊论文

  • 发表日期：

    2026年7月11日

查看全文
  • Maryam Poursaeid,
  • Hadi Seyedarabi,
  • Mohammad Asadpour

摘要

The use of intelligent diagnostic methods for early Alzheimer’s disease detection carries significant implications across various domains, encompassing economic impact and treatment opportunities. This study focuses on Alzheimer’s disease diagnosis and disease progression investigation utilizing transfer learning. Our proposed method employs a transfer learning approach following the acquisition of MRI images from the ADNI database, leveraging the advantage of low computational cost. For preprocessing, the proposed method employs FreeSurfer software, employing a slice-based processing approach along the Axial, Coronal, and Sagittal axes. Three pre-trained models are utilized for feature extraction. Subsequently, an ensemble learning technique combines the strengths of these models across all axes simultaneously. This approach effectively addresses two critical issues in this domain. Upon classifying the three stages of AD, MCI, and CN, the results demonstrate a reasonable diagnostic performance for Alzheimer’s disease. Particularly noteworthy is the classification accuracy of 99.84% along the axial axis in examining disease progression between pMCI and sMCI stages. The presented algorithm facilitates the estimation of disease progression two years prior to Alzheimer’s disease onset without additional computational burden, demonstrating highly competitive accuracy compared to other existing methods.