Optimization-based feature selection in deep learning methods for monkeypox skin lesion detection

This thesis discusses the use of deep learning methods and optimization-based feature selection for the detection of monkeypox skin lesions. Monkeypox disease is an important problem in the health sector today and the development of early diagnosis methods is a great necessity. While traditional diagnostic methods can be costly and time-consuming, deep learning techniques such as CNN offer great potential, especially in detecting skin lesions. In this study, after using the image augmentation method in the data preprocessing steps, features were extracted from the images of skin lesions using the pre-trained model architectures DenseNet-201, ResNet-101 and DarkNet-53. At the same time, optimization methods were used to select meaningful features. Feature selection was carried out with the BGWO method used. The results obtained demonstrate the necessity of the proposed approach to effectively use deep learning methods in detecting skin lesions of monkeypox. It has been determined that the optimization-based feature selection method significantly increases the accuracy of disease diagnosis. This thesis is seen as an important step towards contributing to the development of medical diagnoses and more effective management of patients' health conditions. It is thought that the combination of deep learning and optimization techniques in the early diagnosis of monkeypox can provide great benefits to the healthcare sector.