Diagnosing MonkeyPox from Skin Indications using ArtificialIntelligence Method

Following the first reported case to World Health Organization (WHO) on ۷ May ۲۰۲۲, an ongoing outbreak of monkeypox, a viral zoonosis, has faced public health with a new challenge globally that has shown public health importance on a global scale. In this study, we aim to leverage the promise of Artificial Intelligence (AI) models to diagnose monkeypox disease using images of skin manifestations. An open-access dataset containing skin images of monkeypox, chickenpox, measles, and normal patients was used in this study. After preprocessing and image augmentation, an overall number of ۱۷۵۴ images has been chosen (۸۰% for tarin and ۲۰% for test). Then a modified pre-trained DenseNet-۱۲۱ model was utilized for the four-class classification of mentioned diseases. Achieved results show that the modified pre-trained DenseNet۲۰۱ model offers an average accuracy of ۹۹.۰۴%, precision of ۹۷.۸۵%, sensitivity of ۹۷.۶۶%, specificity of ۹۹.۲۱%, and F۱-Score of ۹۷.۶۱%. Our proposed model offers promising performance for differentiating monkeypox infected patients from chickenpox, measles, and normal patients. Our proposed model can be used in hospitals and clinics to help physicians accurately diagnose monkeypox disease.

کلیدواژه ها:

Monkeypox ، Artificial Intelligence ، Skin Indications