Scholarlink

<p>Pneumonia is a critical respiratory infection that necessitates prompt and accurate diagnosis to prevent severe health complications. Traditional diagnostic methods, such as clinical examination and radiographic analysis, are often time-consuming and subject to inter-observer variability. Recent advances in deep learning have significantly improved the accuracy and efficiency of medical image analysis. In this study, we propose a novel hybrid approach that integrates Capsule Networks, Long Short-Term Memory (LSTM) networks, Generative Adversarial Networks (GANs), and attention mechanisms to enhance pneumonia classification from chest X-ray images. Capsule Networks capture spatial hierarchies, LSTM networks model sequential dependencies, GANs generate high-quality synthetic data to augment the training set, and attention mechanisms improve interpretability and focus. The proposed method was evaluated on a publicly available labeled chest X-ray dataset, achieving an accuracy of 98.65%, precision of 98.5%, recall of 98.5%, F1-score of 99%, and Area Under the Curve of 99%. These results demonstrate the model’s superior performance compared to baseline and state-of-the-art techniques, confirming its potential as a reliable and accurate tool for automated pneumonia diagnosis.</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

Enhanced pneumonia classification from chest X-ray images using hybrid capsule and LSTM networks with attention mechanisms and GAN integration

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • Houmem Slimi,
  • Imen Cherif,
  • Sabeur Abid,
  • Mounir Sayadi

摘要

Pneumonia is a critical respiratory infection that necessitates prompt and accurate diagnosis to prevent severe health complications. Traditional diagnostic methods, such as clinical examination and radiographic analysis, are often time-consuming and subject to inter-observer variability. Recent advances in deep learning have significantly improved the accuracy and efficiency of medical image analysis. In this study, we propose a novel hybrid approach that integrates Capsule Networks, Long Short-Term Memory (LSTM) networks, Generative Adversarial Networks (GANs), and attention mechanisms to enhance pneumonia classification from chest X-ray images. Capsule Networks capture spatial hierarchies, LSTM networks model sequential dependencies, GANs generate high-quality synthetic data to augment the training set, and attention mechanisms improve interpretability and focus. The proposed method was evaluated on a publicly available labeled chest X-ray dataset, achieving an accuracy of 98.65%, precision of 98.5%, recall of 98.5%, F1-score of 99%, and Area Under the Curve of 99%. These results demonstrate the model’s superior performance compared to baseline and state-of-the-art techniques, confirming its potential as a reliable and accurate tool for automated pneumonia diagnosis.