Cervical Cancer Detection Using Support Vector Machine

The pre-cancer can be identified by most efficient method of segmentation and by classification in a pap
smear images. The cervical cell cancer can be detected by monitoring exactly the changes in a cell .The
major challenging task is to identify the overlapping of cytoplasm which is accurately done in this work.
The earlier cervical cancer is diagnosed, the more successfully it can be treated. Regular cervical
screening can save thousands of live every year. The majority of these deaths could be prevented if all
women had undergone cervical screening. From the global survey say a female should start screening at
the age of 21, or within 3 years of her first sexual encounter - whichever occurs first. Cervical screening
does not detect cancer, it simply looks for abnormal changes in the cells of the cervix. If left untreated,
some abnormal cells can eventually develop into cancer. The input is fetched from the dataset and those
images been processed using efficient function of resizing, grayscale conversion and noise reduction. Highlevel shape information to guide segmentation where cell boundary might be weak or lost due to cell
overlapping. In the segmentation Canny edge detection model used to detect the feature. The feature is
extracted accurately through histogram of gradient. The entire classification is done by support vector
machine. An evaluation carried out using two different datasets ISBI 2015 dataset and SZU dataset of the
proposed method over the state-of-the-art methods in terms of classification accuracy