Classification Of Cervical Cancer Stage From Pap Smear Tests

This research focuses on the field of biomedical engineering in the works of Pap smear
image analysis. Pap smear test is an efficient procedure in detecting cases of cervical
cancer especially in early stages. However, most of these tests are done manually by
medical personnel, which remains a tedious task to carry out on a daily basis due to
the occurrence of human and technical error. The purpose of this research is to identify
an effective algorithm to classify the presence of abnormalities in the given Pap smear
samples. The proposed approach will implement stages of image pre-processing,
feature selection and extraction as well as classification of classes. During image preprocessing,
the image will be converted to greyscale before improving their contrast
level for better analysis. Feature extraction is then used to select the appropriate
features that contribute most to the predicted variable from the image. Then,
classification methods for the classification of classes in these cells such as K-Nearest
Neighborhood (KNN) and Support Vector Machine (SVM) were explored. The
performance of the proposed classification algorithm gave satisfactory results of
accuracy, 91.9% for KNN classification and 95.0% for SVM classification.