Computer aided approach for cervical cancer detection and classification using optimization

Abstract

Cancer is a life threatening disease. Majority of deaths occur due to cancer. Breast and Cervical cancers are the most live killing diseases of women patients in the world. Breast cancer occurs externally in the women patients and it can be detected by scanning the affected regions. The women patients who are affected by breast cancer can check themselves. Hence it can be detected in women patients at an early stage. The cervical cancer attacks womens cervix internally and it can be detected by scanning the internal region of the vagina. Human Papilloma Virus (HPV) is the main cause of cervical cancer formation in women patients. This virus initially affects the cells in the cervical region of the women patients and spreads over the entire region of the cervix.The women patients who are affected by cervical cancer, unable to check themselves; hence it cannot be detected at an earlier stage. The main reason for the death due to cervical cancer in women patients is that it cannot be detected at an earlier stage and the patients can not receive any symptoms until they reach final stage of the cancer. The death ratio of the women patients can be reduced if it is detected at an earlier stage. Hence, this research proposes a methodology to detect the cervical cancer at an earlier stage to prevent death of women patients. Automated detection and classification procedures are presented for detection of cancer from the cervical images using clinically significant and biologically interpretable set of features. The original cervical image is initially preprocessed, in which the image is applied with image enhancement technique using Oriented Local Histogram Equalization (OLHE) and then the enhanced image is transformed into Dual Tree Complex Wavelet Transform. The preprocessed image is then used to extract features such as wavelet features, Grey Level Co-occurrence Matrix (GLCM) features and moment invariant features. These features are used to train the neural network classifier to classify the cervical image into benign