Aplicación de Deep Learning como inteligencia artificial para el diagnóstico del cáncer de cuello uterino en citología en base líquida en comparación con técnicas tradicionales: revisión sistemática

Abstract

Introduction Cervical cancer remains one of the main cancerous pathologies that causes death in women worldwide, especially in countries with scarce resources. Conventional cytology and histopathological analysis, although irreplaceable, are often based on the expert’s personal opinion and on technical equipment that is not always available. In this context, AI and, particularly, deep learning, is emerging as a profitable alternative to enhance health diagnostic processes. Objective: To analyze the impact of AI applications on the accuracy and effectiveness of histopathological diagnosis of cervical cancer through a systematic review of recent literature. Methodology: A systematic review was conducted following the PRISMA 2020 guidelines. The review covered scientific articles published from 2020 to 2025, retrieved from databases such as PubMed, Scopus, and Cochrane. Diagnostic cytology and histopathological case studies using deep learning techniques, particularly convolutional neural networks, were included in the review. The Rayyan platform assisted in record management, while QUADAS-2 was used to assess the risk of bias. Sensitivity, specificity, and clinical feasibility were considered as the main parameters. Expected Results: It is expected to identify AI methodologies with high diagnostic accuracy, as well as to demonstrate the challenges involved in their implementation in real clinical settings. The intention is to promote improvements in early detection systems and reduce health inequalities, particularly in resource-limited regions.