Scholarlink

<p>Pneumonia continues to be one of the global causes of morbidity and mortality, necessitating early and proper diagnosis using chest X-rays (CXRs). Deep learning networks, specifically convolutional neural networks (CNNs), have shown excellent results in identifying pneumonia. But their performance is normally hampered by dataset imbalance, feature redundancy, and local receptive fields, reducing generalizability. To overcome such challenges, the study introduces Vi-GeN 1.0, a GAN-augmented Vision Transformer (ViT) pipeline specifically architected for pneumonia classification robustness. Our method uses Wasserstein GAN with Gradient Penalty (WGAN-GP) to synthesize high-quality CXRs, improving dataset diversity and feature learning. The ViT classifier, trained on the real and synthetic data, learns global contextual representations, resulting in better classification performance. Vi-GeN 1.0 model was tested on the Kaggle Chest X-ray Pneumonia dataset, showing 97.3% accuracy, 98.1% sensitivity, 96.5% specificity, and AUC-ROC of 0.985, performing much better than the baseline ViT model (<i>p</i> &lt; 0.01). The realism and feature preservation of the synthetic images are ensured by FID (22.1), SSIM (0.87), and PSNR (28.9&#xa0;dB) validation. Our results showed that GAN-based augmentation improves pneumonia classification performance effectively, making Vi-GeN 1.0 a valuable candidate for clinical use, especially in low-resource environments. Future research will investigate multi-class classification, domain generalization, and more sophisticated augmentation methods like diffusion models.</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

Vi-GeN 1.0: GAN-Augmented Vision Transformer Pipeline for Diversified Pneumonia Classification

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月18日

查看全文
  • Archana Dash,
  • Soumyarashmi Panigrahi,
  • Debasish Swapnesh Kumar Nayak,
  • Amiya Prasad Dash,
  • Tripti Swarnkar

摘要

Pneumonia continues to be one of the global causes of morbidity and mortality, necessitating early and proper diagnosis using chest X-rays (CXRs). Deep learning networks, specifically convolutional neural networks (CNNs), have shown excellent results in identifying pneumonia. But their performance is normally hampered by dataset imbalance, feature redundancy, and local receptive fields, reducing generalizability. To overcome such challenges, the study introduces Vi-GeN 1.0, a GAN-augmented Vision Transformer (ViT) pipeline specifically architected for pneumonia classification robustness. Our method uses Wasserstein GAN with Gradient Penalty (WGAN-GP) to synthesize high-quality CXRs, improving dataset diversity and feature learning. The ViT classifier, trained on the real and synthetic data, learns global contextual representations, resulting in better classification performance. Vi-GeN 1.0 model was tested on the Kaggle Chest X-ray Pneumonia dataset, showing 97.3% accuracy, 98.1% sensitivity, 96.5% specificity, and AUC-ROC of 0.985, performing much better than the baseline ViT model (p < 0.01). The realism and feature preservation of the synthetic images are ensured by FID (22.1), SSIM (0.87), and PSNR (28.9 dB) validation. Our results showed that GAN-based augmentation improves pneumonia classification performance effectively, making Vi-GeN 1.0 a valuable candidate for clinical use, especially in low-resource environments. Future research will investigate multi-class classification, domain generalization, and more sophisticated augmentation methods like diffusion models.