Effective Classification of Cervical Cancer Using Contourlet Transform
Cervical cancer is the second most leading cancer among women. Source of cervical cancer is Human Papilloma Virus (HPV). Tests to confirm a diagnosis of cervical cancer are colposcopy and cervical biopsy (pap smear). This paper includes the usage of colposcopy images and tests to find where cancer cells are on the surface of the cervix. Contourlet transform (CT) is proposed to extract the features of the colposcopy images and then for classification of Aceto white Region (Abnormal) and Non Aceto white Region (Normal), K-Nearest Neighbourhood (KNN) classifier is used. In the existing system, wavelet transforms are used to extract the features in which minimum quantity of information and features are obtained (Only 3 directions are focused). In CT, for higher resolution, additional directional is employed. The greater performance is achieved by extracting and choosing the simplest features from contourlet coefficients of the colposcopy images and these outputs are fed into the KNN classifier for classification.