The second most frequent malignancy in women globally is cervical cancer, with a 60% fatality rate, which arises from the change of cells in the female cervix, and is one of the treatable malignancies when detected in its early stages. Thus, the goal of early detection of cervical cancer is to diminish the mortality rate. Unfortunately, the diagnosing procedure is inefficient and imprecise because it relies primarily on the pathologist's experience. Considering this issue, the researchers have started working on the automatic diagnosis of cervical cancer for preventing the misclassification of cancerous and noncancerous cells. This review paper presents various deep learning approaches for the automatic diagnosis of cervical cancer which is more accurate than the traditional approach. Moreover, weaknesses, strengths, accuracy, other performance metrics, and dataset description have been highlighted for each respective technique. The paper also addresses the classification, segmentation, and feature extraction that will help pathologists with an efficient diagnosis process. The survey examines nine years' worth of articles, conference papers, and journals on deep learning in the automated diagnosis and categorization of cervical cancer. Moreover, the study examines 55 papers collected via electronic means from reputed scientific databases.

Cancer, Deep Learning, Pre-Cancerous, Segmentation, Virus