Identification of Cervical Cancer Cells using Backpropagation

Abstract Cervical cancer is the growth of abnormal cells in the cervix. In general, cervical cancer can be early detected through Pap smear. This examination aims to see the existence of changes in cell shape from digital microscopic images, but it is still manually performed. Therefore, a method is required to identify the cell of cervical cancer automatically. Under this research, the proposed method consists of three primary processes. The initial stage is image input. The process continues to the pre-processing phase which involves scaling, grey-scaling, median filter and thresholding. The final step is identification process using Backpropagation to determine the normal and abnormal cells of cervical cancer. In this research, the training and testing processes used 40 data and 20 data respectively. According to the conducted test, 18 data were correctly identified as the expected output. The study concluded the proposed method performs well in identifying the cervical cancer cells with an accuracy rate of 90%.