CERVISCAN: CERVICAL CANCER DETECTION USING CONVOLUTIONAL NEURAL NETWORK (CNN)
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Malaysia Sarawak
Abstract
Description
Cervical cancer remains a major factor in cancer-related deaths among women in the world, especially in areas with limited healthcare resources. The current screening methods, like Pap smear tests, take time and demand expertise and resources due to the need for manual identification of abnormal cervical cells. However, cervical cancer has the potential to be treated if it is detected in the early phases. Therefore, this project, CerviScan, proposes an automated cervical cancer detection system integrating a Convolutional Neural Network (CNN) model trained on the publicly accessible SIPaKMeD Dataset. The CNN model utilises transfer learning with ResNet-50 to classify the cervical Pap smear images into five categories, such as Dyskeratotic, Koilocytotic, Metaplastic, Parabasal, and Superficial-Intermediate. Metrics like accuracy, precision, recall, and F1-score are used to evaluate the model’s performance and effectiveness. Moreover, a web-based system that integrates this model is developed to allow medical practitioners to upload Pap smear images, generate and download classification results, as well as manage the list of patients. Last but not least, by reducing the dependency on manual analysis of abnormal cervical cells, CerviScan aims to enhance early detection of cervical cancer, particularly in low-resource settings. Thus, improving the clinical efficiency and treatment success rates.