Computer Aided Screening of Cervical Cancer Using Random Forest Classifier

Cervical cancer is a pathological disorder of the human that affects around 50 million people worldwide. The early detection and diagnosis of cervical cancer is important for the women to prevent it in an earlier stage. Conventional approaches used abnormal lesion size and shape to detect the cancer area, which provides low accuracy for surgery. In this paper, computer aided automatic cervical cancer detection methodology is proposed to overcome the limitations of the conventional methods. In preprocessing stage, complex wavelet transform is used to convert the time domain image into multi resolution image and further features are extracted from this transformed wavelet coefficients. Then, the random forest classifier is used to train and test the cervical cancer.