Monkeypox Skin Lesion Detection with MobileNetV2 and VGGNet Models


Irmak M. C., Aydın T., Yağanoğlu M.

2022 Medical Technologies Congress (TIPTEKNO), Antalya, Türkiye, 31 Ekim - 02 Aralık 2022, ss.1-4

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/tiptekno56568.2022.9960194
  • Basıldığı Şehir: Antalya
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.1-4
  • Van Yüzüncü Yıl Üniversitesi Adresli: Evet

Özet

One of the viral diseases that has caused concern in many countries after the COVID-19 pandemic is monkeypox virus. To date, outbreaks have been reported in 75 countries. Monkeypox is difficult to diagnose at an early stage because its symptoms in the human body are similar to those of chickenpox and measles. Because this virus was a rare disease before the current epidemic, it has created an information gap among health professionals. It is thought that computer-aided detection methods will be useful in cases where the polymerase chain reaction (PCR) tests needed to diagnose the disease are not yet available. Recently, many diseases, including COVID-19, have been successfully detected by deep learning methods after sufficient images were available. In this study, classification was performed using the previously trained CNN networks MobileNetV2, VGG16, and VGG19 on the Monkeypox Skin Image Dataset, which was made open source in 2022, and the accuracy metrics of these three methods were compared. The highest performance scores were obtained with MobileNetV2, with 91.38% accuracy, 90.5% precision, 86.75% recall and 88.25% f1 score. The VGG16 method achieved 83.62% accuracy and the VGG19 method achieved 78.45% accuracy.