Segmentation of the Cervix in Colposcopy Images Using Machine Learning Techniques

Cervical cancer caused by human Papillomavirus (HPV), a sexually transmitted disease, is one of the most common neoplasms in women nationally and globally. Although there are health campaigns promoting screening tests to detect the disease, the waiting times for results are high due to deficiencies in laboratory infrastructure, affecting diagnosis. In this work, we propose to apply machine learning techniques for cervical segmentation in colposcopy images obtained during cytology, more specifically the cervix region for supporting a further classification stage. Finally, we develop a desktop application only for unsupervised learning models. Also, this work is a result of the project CITOBOT, funded by Minciencias and developed by a multidisciplinary team. Results show acceptable metrics in the segmentation of the cervix, both in unsupervised and supervised methods.