TAF PUBLISHING

Cervical cancer is a kind of cancer disease caused by human papillomavirus types 16 and 18 attacking women's cervix. To detect cervical cancer, the frequently-used method is Pap-Smear; however, errors often occur when the method is taken to diagnose the level of cervical cancer. Thus, a proper system is required, which is supposed to help identify the result of Pap-Smear. This study aims at designing a system to detect the symptoms of cervical cancer using MATLAB to solve these errors. The image processing begins with converting the type of image, followed by thresholding and noise removal using filters until the image has become ready to be detected. For a thresholding process, an Adaptive Thresholding method is taken, in which the thresholding focuses on local threshold values. The system can classify images into two types, i.e., normal and abnormal (precancerous). Abnormal type is divided into three subtypes, i.e., mild, moderate, and severe. An experiment is conducted on the proposed system, in which it is supposed to analyze 500 test images, including 250 for training and 250 for testing. A perfect 100% accuracy rate is obtained based on the testing process, while the average processing time is 25.4 seconds with a WS value at ten and a C value at -2.