EXPLORING THE IMPACT OF INCREASING THE NUMBER OF CLASSES ON THE PERFORMANCE OF CERVICAL CANCER DETECTION MODELS USING DEEP LEARNING AND COLPOSCOPY

New technologies, specifically employing deep learning algorithms, have resulted in the development of enhanced expeditious and accurate cervical cancer screening, particularly within the medical field. The colposcopy approach is one of the crucial parts of clinical screening for early detection of cervical cancer and cervical intraepithelial neoplasia (CIN). It has an immediate impact on the diagnosis and treatment plan for the patient. Therefore, it has potential as it has been extensively used for cervical cancer screening. The present work proposes a predictive model implementing Convolutional Neural Networks (CNN) and colposcopy images for the identification of various classifications of cervical cancer, encompassing its initial stages. The investigation looks into the implications of augmenting the total number of classes on the proposed model's precision. This presents a prospect for the different-stage diagnosis of cervical cancer. This presents a prospect for the early-stage diagnosis of cervical cancer. The findings indicate that according to the results, augmenting the number of classes enhances detection accuracy during the training phase, reaching up to diminishing to 43.11% in the proposed model. For future research, the application promise for enhancing the detection of cervical cancer using colposcopy images.