Neural Network to detect Cervical Cancer early using Pap smears Images
Mithlesh Arya, Praveen Kumar Yadav, A. C. JAIN, Megha Gupta, Saroj Agarwal, Arvind Singh Rajpoot
Abstract
In India uterine cervix cancer is the moment most common cancer among ladies. Cancer of Uterine Cervix in rural area women are at higher risk than the urban area women due to less awareness. Cancer of Uterine Cervix is also known as cervical cancer. The Pap spread may be a screening test utilized to identify cervical cancer. Pap smear screenings are conducted by pathologists, but there is a shortage of pathologists in India. Automatic detection method helps in the early detection of cervical cancer. Based on the literature review, segmentation, feature extraction and classification are used for classification of cervical cancer cells. In this paper Convolution Neural Network is used for the classification of the datasets into 2 classes (normal and abnormal). In this paper, Pap smear images are collected from Pathology Lab at Jaipur. First the CNN model was applied on the public available dataset and for the validation of work it was implemented on PapsmearJP dataset. Several epochs are employed for the different ratio of training and testing dataset on public dataset SIPaKMeD and PapsmearJP dataset. In this paper, we achieved a maximum accuracy of $\mathbf{9 6 \%}$ with 70 epochs on $\mathbf{8 0 \%}$ training and $\mathbf{2 0 \%}$ testing dataset on PapsmearJP dataset.