Scholarlink

<p>Deep learning (DL) has emerged as a transformative technology in medical image diagnosis, surpassing traditional machine learning algorithms in handling complex and large datasets. This paper presents a comprehensive review of state-of-the-art DL models, including 3D-2D convolutional neural networks (CNNs), Multimodal CNN, recurrent neural networks, long short-term memory, gated recurrent unit, and transfer learning. We also explore advanced architectures such as U-Net and vision transformers, discussing their principles, advantages, and limitations. The review focuses on applications in segmentation, classification, detection, and localization, with a particular emphasis on medical image diagnosis. Here, we show that these models have achieved remarkable performance, exemplified by accuracy rates exceeding 95% in MRI brain tissue segmentation and 94% in Alzheimer's disease diagnosis. By analyzing these models, we aim to provide insights into their effectiveness and potential for future advancements in medical imaging.</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

Innovative deep learning architectures for medical image diagnosis: a comprehensive review of convolutional, recurrent, and transformer models

  • 期刊论文

  • 发表日期：

    2025年8月5日

查看全文
  • Fathia Ghribi,
  • Fayçal Hamdaoui

摘要

Deep learning (DL) has emerged as a transformative technology in medical image diagnosis, surpassing traditional machine learning algorithms in handling complex and large datasets. This paper presents a comprehensive review of state-of-the-art DL models, including 3D-2D convolutional neural networks (CNNs), Multimodal CNN, recurrent neural networks, long short-term memory, gated recurrent unit, and transfer learning. We also explore advanced architectures such as U-Net and vision transformers, discussing their principles, advantages, and limitations. The review focuses on applications in segmentation, classification, detection, and localization, with a particular emphasis on medical image diagnosis. Here, we show that these models have achieved remarkable performance, exemplified by accuracy rates exceeding 95% in MRI brain tissue segmentation and 94% in Alzheimer's disease diagnosis. By analyzing these models, we aim to provide insights into their effectiveness and potential for future advancements in medical imaging.