Scholarlink

<p>Nail disease classification is a crucial task in dermatology, aiding in the early diagnosis and treatment of various conditions. In this study, we leverage an open-access dataset from Kaggle containing 3835 images and apply data augmentation techniques, expanding the dataset to 11,505 images to improve model generalization. We propose a CNN-based deep learning model and evaluate its performance on the augmented dataset. To further enhance classification accuracy, we fuse the proposed CNN model with a Capsule Network (CapsNet), leveraging its ability to capture spatial hierarchies and complex relationships between image features. Both models are trained and evaluated, followed by a visualization of classification results. The fused CNN–CapsNet model outperforms the standalone CNN model, achieving an overall accuracy of 98.5%, demonstrating precise and secure AI-powered nail disease diagnosis, ensuring model robustness. This research highlights the advantages of combining CNNs with Capsule Networks for improved medical image analysis and classification.</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

AI-powered precise diagnosis and automated nail disease detection using a fused CNN–CapsNet model

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月3日

查看全文
  • Vatsala Anand,
  • Ajay Khajuria,
  • Mohammed Shuaib,
  • Irfanullah Khan,
  • Shadab Alam,
  • Mehran Ullah

摘要

Nail disease classification is a crucial task in dermatology, aiding in the early diagnosis and treatment of various conditions. In this study, we leverage an open-access dataset from Kaggle containing 3835 images and apply data augmentation techniques, expanding the dataset to 11,505 images to improve model generalization. We propose a CNN-based deep learning model and evaluate its performance on the augmented dataset. To further enhance classification accuracy, we fuse the proposed CNN model with a Capsule Network (CapsNet), leveraging its ability to capture spatial hierarchies and complex relationships between image features. Both models are trained and evaluated, followed by a visualization of classification results. The fused CNN–CapsNet model outperforms the standalone CNN model, achieving an overall accuracy of 98.5%, demonstrating precise and secure AI-powered nail disease diagnosis, ensuring model robustness. This research highlights the advantages of combining CNNs with Capsule Networks for improved medical image analysis and classification.