Scholarlink

<p>Melanoma remains one of the most challenging skin cancers to detect due to its visual similarity with benign lesions. This study proposes a spatial mean-based feature fusion and ensemble learning framework for automated melanoma detection. The approach integrates feature embeddings from multiple pre-trained deep models through spatial mean fusion, producing a unified representation that preserves spatial and contextual information while reducing redundancy. The fused features are normalized and classified using an ensemble of machine learning models, including Random Forest, Support Vector Machine (SVM), XGBoost, Decision Tree and Logistic Regression. Experiments on a benchmark skin lesion dataset confirm that the proposed framework achieves higher accuracy, precision, recall, and F1-score compared to single classifiers and conventional ensemble strategies. The results demonstrate the effectiveness of the proposed approach, establishing its potential as a reliable solution for automated melanoma diagnosis. The source code is available at: <a href="https://shorturl.at/Gq0cS">https://shorturl.at/Gq0cS</a>, and the dataset can be accessed at: <a href="https://www.kaggle.com/datasets/hasnainjaved/melanoma-skin-cancer-dataset-of-10000-images">https://www.kaggle.com/datasets/hasnainjaved/melanoma-skin-cancer-dataset-of-10000-images</a>.</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

A spatial mean-based feature fusion and ensemble learning framework for enhanced melanoma detection

  • 期刊论文

  • 发表日期：

    2025年10月25日

查看全文
  • Brahmanand Dubey,
  • Anubhav Dwivedi,
  • Subhayu Ghosh,
  • Nanda Dulal Jana

摘要

Melanoma remains one of the most challenging skin cancers to detect due to its visual similarity with benign lesions. This study proposes a spatial mean-based feature fusion and ensemble learning framework for automated melanoma detection. The approach integrates feature embeddings from multiple pre-trained deep models through spatial mean fusion, producing a unified representation that preserves spatial and contextual information while reducing redundancy. The fused features are normalized and classified using an ensemble of machine learning models, including Random Forest, Support Vector Machine (SVM), XGBoost, Decision Tree and Logistic Regression. Experiments on a benchmark skin lesion dataset confirm that the proposed framework achieves higher accuracy, precision, recall, and F1-score compared to single classifiers and conventional ensemble strategies. The results demonstrate the effectiveness of the proposed approach, establishing its potential as a reliable solution for automated melanoma diagnosis. The source code is available at: https://shorturl.at/Gq0cS, and the dataset can be accessed at: https://www.kaggle.com/datasets/hasnainjaved/melanoma-skin-cancer-dataset-of-10000-images.