Prediction of multiclass cervical cancer using deep machine learning algorithms in healthcare services
Sanda Shakti Kharisma Jaya, M. Latha · 2021
Cervical cancer is the least unsafe tumor that develops in a woman’s abdomen. Cervical cancer is a two-form disease, normal condition and abnormal condition. Both show similar patterns of nucleus and cytoplasm. There are seven stages in a cervical cancer, which includes superficial squamous, intermediate squamous, columnar epithelial, mild dysplastic, moderate dysplastic, severe dysplastic, and carcinoma in situ. The prediction of a cervical cancer stage is very difficult, leave alone if a person is affected by tumor or not. In such a case, statistical analytics and machine learning algorithms are very helpful to predict the state of the cancer in early stages. The main focus of this chapter is in analyzing the various deep machine learning algorithms and finding the best performance measure that gives the maximum accuracy level for prediction of cervical cancer. Statistical models of size, shape, and structure are most important for feature extraction from Pap smear images. From the dataset, 237 samples were collected and used for developing an applied mathematics model to analyze the texture variation and study the correlation between shape and texture. Various classifiers were implemented and compared under machine learning algorithms to measure the accuracy level. This proposed article has been analyzed from three kinds of environments, such as data analytics, big data, and digital image processing. The entire execution of this chapter has been implemented using MATLAB R2016a.