Segmentation and Classification Techniques for Pap Smear Images in Detecting Cervical Cancer: A Systematic Review

Cervical cancer is the fourth most common disease and a significant concern for women’s health worldwide. This systematic review explored the segmentation and classification of Pap smear images for cervical cancer detection using deep learning techniques in articles published between 2016 and 2023. The Pap smear test was used to screen the internal cervical organs for abnormal cells. Researchers have developed various advanced image processing approaches, feature extraction methods, and classification algorithms to detect cervical cancer by characterizing the contents of Pap-smeared images. This review aimed to provide a comprehensive assessment of the most recent techniques in the scientific field by presenting a review of published articles related to cervical smear screening. The two key focal points of this review are the cell image segmentation and classification algorithms for categorizing Pap smear images for cervical cancer detection. Among the selected studies, Unet and Mask RCNN stand out as the most widely used techniques for segmenting Pap smear images, while SVM and CNN are used to classify their levels based on the extracted features. Most of the algorithms achieved more than 90% accuracy on the Pap smear images. For binary classification, accuracies of 98.32% and 98.86% were achieved for the Herlev and SIPaKMeD datasets, respectively, using a hybrid feature fusion technique. The effective utilization of deep learning techniques may have the potential to enhance the accuracy and efficiency of cervical cancer detection in early diagnosis.