EXTRACTING COLOUR FEATURES FROM COLPOSCOPY IMAGES FOR CLASSIFICATION OF CERVICAL CANCER
Loading...
Files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Malaysia Sarawak (UNIMAS)
Abstract
Description
Cervical cancer is a major public health issue that affects millions of women worldwide, highlighting the critical importance of early detection and diagnosis. This project investigates the application of colour feature extraction from colposcopy images to improve the classification accuracy of cervical cancer using machine learning techniques. The research begins with the preprocessing of colposcopy images to improve their quality, which includes resizing all images to a uniform dimension to ensure consistency during colour feature extraction. Colour features are extracted using a variety of methods, including the computation of dominant, mean, and standard deviation values in the Red, Green, Blue (RGB) colour space, enabling a comprehensive analysis of the colour distributions present in the images. The effectiveness to differentiate between normal and abnormal cervical tissues is analysed using machine learning algorithms such as Support Vector Machine (SVM), Random Forest, eXtreme Gradient Boosting (XGBoost), Decision Tree and Logistic Regression. The findings of this study show a significant improvement in classification accuracy, with the extracted colour features providing useful information for the early diagnosis of cervical cancer. Furthermore, the study highlights the potential of automated image processing approaches to help doctors make better diagnostic decisions, ultimately aiming to improve patient outcomes through timely and accurate classification of cervical abnormalities. This study lays a foundation for future developments in automated cervical cancer screening tools, emphasising the integration of colour feature analysis into medical imaging and its value in improving healthcare solutions.