Cervical cancer detection method using an improved cellular neural network (CNN) algorithm

Abstract

Cervical cancer is the second most common in Malaysia and the fourth frequent cancer among women in worldwide.  Pap smear test is often ignored although it is actually useful, beneficial and essential as screening tool for cervical cancer. However, Pap smear images have low sensitivity as well as specificity. Therefore, it is difficult to determine whether the abnormal cells are cancerous or not. Recently, computer-based algorithms are widely used in cervical cancer screening. In this study, an improved cellular neural network (CNN) algorithm is proposed as the solution to detect the cancerous cells in real-time by undergoing the image processing of Pap smear images. A few templates are combined and modified to form an ideal CNN algorithm to detect the cancerous cells in total of 115 Pap smear images. A MATLAB based CNN is developed for an automated detection of cervix cancerous cells where the templates segmented the nucleus of the cells. From the simulation results, our proposed CNN algorithm can detect the cervix cancer cells automatically with more than 88% accuracy.