Scholarlink

<p>The detection of the Monkeypox virus outside of Africa, with its identification in Sweden in August, followed by the World Health Organization's declaration of a global health emergency, has heightened concerns about the potential emergence of a new epidemic. Effectively managing disease detection and isolation processes is crucial to avoid the adverse impacts of a global pandemic. A hybrid classification-based approach is presented for detecting Monkeypox, focusing on the early symptoms of skin lesions, which typically appear within three days. Due to the limited availability of data, Convolutional Neural Network models were employed. The proposed method integrates multiple Convolutional Neural Network architectures, including MobileNetV3Large and EfficientNetB0, to enhance the effectiveness of feature extraction. Classification was performed using an optimized Support Vector Machine method, refined through Grid Search to ensure optimal parameter selection. The optimized Support Vector Machine highly improved the classification accuracy. This approach, aimed at generalizability, has demonstrated successful results on the Monkeypox Skin Images Dataset and Monkeypox Skin Lesion Dataset open-access datasets. The proposed method achieved a maximum accuracy of 98.67% on the Monkeypox Skin Images Dataset and 98.13% on the Monkeypox Skin Lesion Dataset when cross-validation was applied to the generated dataset partitions. This study is considered a significant step toward the early detection of Monkeypox and the development of effective intervention strategies, which will contribute to preventing future outbreaks.</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

Efficient monkeypox detection using hybrid lightweight CNN architectures and optimized SVM with grid search on imbalanced data

  • 期刊论文

  • 发表日期：

    2025年2月24日

查看全文
  • Dilber Çetintaş

摘要

The detection of the Monkeypox virus outside of Africa, with its identification in Sweden in August, followed by the World Health Organization's declaration of a global health emergency, has heightened concerns about the potential emergence of a new epidemic. Effectively managing disease detection and isolation processes is crucial to avoid the adverse impacts of a global pandemic. A hybrid classification-based approach is presented for detecting Monkeypox, focusing on the early symptoms of skin lesions, which typically appear within three days. Due to the limited availability of data, Convolutional Neural Network models were employed. The proposed method integrates multiple Convolutional Neural Network architectures, including MobileNetV3Large and EfficientNetB0, to enhance the effectiveness of feature extraction. Classification was performed using an optimized Support Vector Machine method, refined through Grid Search to ensure optimal parameter selection. The optimized Support Vector Machine highly improved the classification accuracy. This approach, aimed at generalizability, has demonstrated successful results on the Monkeypox Skin Images Dataset and Monkeypox Skin Lesion Dataset open-access datasets. The proposed method achieved a maximum accuracy of 98.67% on the Monkeypox Skin Images Dataset and 98.13% on the Monkeypox Skin Lesion Dataset when cross-validation was applied to the generated dataset partitions. This study is considered a significant step toward the early detection of Monkeypox and the development of effective intervention strategies, which will contribute to preventing future outbreaks.