Deep Learning based Classification of Cervical Cancer using Transfer Learning

Cervical cancer is responsible for 90% of all deaths in low- and middle-income nations (LMIC). Cervical cancer is still one of the most common cancers in women's that develops in the cervix lower part of the womb that connects to the vagina canal. Cervical cancers are caused by continual infection on their cervix with one of the human papilloma viruses (HPVs). The most commonly used screening test for early detection of abnormal cells and cancer is the Pap smear test. Manual screening, on the other hand, leads to human errors. It is possible to save lives by detecting cancer early and accurately. Transfer learning has made significant progress in the field of machine learning in recent years, and the use of transfer learning technology to cervical cancer image classification has emerged as a new experimental domain. This work presents a study of transfer learning frameworks InceptionResNetV2, VGG19, DenseNet201 and Xception networks pre-trained on ImageNet, to classify cervical images using the SIPaKMeD dataset.