Classification of human monkeypox with the Fuzzy C-Means Algorithm using image processing methods and Haralick texture parameters

Authors : Senem Gönenç, Ozge Pasin

Pages : 82-92

Doi:10.21673/anadoluklin.1477313

View : 39 | Download : 33

Publication Date : 2025-01-29

Article Type : Research Paper

Abstract :Aim: Human monkeypox can cause skin lesions in the form of blisters of different shapes on various parts of the body. Due to the fact that the skin lesions caused by human monkeypox have a very similar appearance to lesions caused by chickenpox and measles, the study includes images of chickenpox and measles as well as images of human monkeypox. The aim of this study is to distinguish human monkeypox virus skin lesion images from other viral diseases with similar images. Methods: For this study, the Monkeypox Skin Lesion Dataset, which consists of binary classification data for monkeypox and non-monkeypox (chickenpox, measles) skin lesions, is accessed from the Kaggle.com website. In total, 228 images are processed, with 101 images in the monkeypox group and 127 images in the non-monkeypox group. The images in the Monkeypox Skin Lesion Dataset are processed using image analysis methods and Haralick texture parameters are calculated to create 13 different features for each image. For the classification process in the statistical analysis part of the study, Fuzzy C-Means algorithm is used. Results: The images used in the study belong to individuals with varying skin tones and from different parts of the body, and the algorithm provides encouraging results in determining the type of skin lesions in the images. The overall classification accuracy rate is 61.8%, and the highest accuracy (76.2%) is achieved in the monkeypox class. Conclusion: This study demonstrates that images of viral diseases with similar skin lesions can be classified using various image-processing techniques and different statistical methods.
Keywords : Gruplama, maymun poks, sınıflandırma

ORIGINAL ARTICLE URL