Scholarlink

<p>Skin melanoma cases have increased significantly in recent years, partly due to environmental changes. Early detection of skin melanoma helps patients receive timely treatment. While deep learning methods have been used for melanoma detection, but there is still scope for improvement. In this study, a combined approach based on three different deep learning techniques is developed to detect skin melanoma more effectively. The skin images are first processed using a black-hat transformation to remove hair. Both individual deep learning methods and an ensemble of neural networks are applied for melanoma detection. The method’s performance is evaluated using metrics like accuracy, recall, precision, and F1-score. The proposed ensemble method shows better results compared to existing approaches, achieving an accuracy of 98.52%. This method can assist doctors in making more accurate diagnoses and providing a reliable second opinion.</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 Skin Melanoma Detection Technique Based on Black-Hat Transformation and Ensemble of Deep Learning Models

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Naliniprava Behera,
  • Akhilendra Pratap Singh,
  • Jitendra Kumar Rout,
  • Bunil Kumar Balabantaray,
  • Sambit Kumar Mishra

摘要

Skin melanoma cases have increased significantly in recent years, partly due to environmental changes. Early detection of skin melanoma helps patients receive timely treatment. While deep learning methods have been used for melanoma detection, but there is still scope for improvement. In this study, a combined approach based on three different deep learning techniques is developed to detect skin melanoma more effectively. The skin images are first processed using a black-hat transformation to remove hair. Both individual deep learning methods and an ensemble of neural networks are applied for melanoma detection. The method’s performance is evaluated using metrics like accuracy, recall, precision, and F1-score. The proposed ensemble method shows better results compared to existing approaches, achieving an accuracy of 98.52%. This method can assist doctors in making more accurate diagnoses and providing a reliable second opinion.