A Novel approach for indentifying cervical cancer using enhanced segmentation and classification technique

Abstract

Cervical malignancy is a disease, which is generally caused by virus named Human Papilloma Virus (HPV). The world health organization survey says there are around 5,30,000 new cases of cervical cancer enlisted worldwide. The disease can be cured if diagnosed in the early stage. The different causes for the disease are early sexual contact, multiple sexual partners, and conception of prevention pills. The different symptoms of cervical cancer are weight loss, fatigue, back pain, leg pain, leakage of urine, bleeding between periods, and abnormal changes in cervix. Pap test, colposcopy is widely used to inspect the cervix and the vagina. The test identifies irregular cells in the cervix, and they are classified from normal to abnormal. If the test result is positive, it will be infected with cervical cancer. The Pap test has several disadvantages including expertise dependent, low sensitivity, and frequent retesting. From this point, the need for cervical image segmentation begins. Medical images obtained from the Pap test are segmented into multiple objects using different techniques like thresholding, level set, active contour, k-means, watershed, and morphological operations. The major problem is image degradations such as blur, noise, and color or contrast imperfection which severely affect the segmentation process. The pap test has several disadvantages including expertise dependent, low sensitivity, and frequent retesting. Segmentation mainly focuses on separating nucleus from the background region. Classification algorithm is used for classifying cervix images into normal or suspicious. newline