Cérvix tool: Detección y clasificación temprana de neoplasia intraepitelial cervical (NIC) a través de IA.

Abstract

Abstract: Cervical cancer remains a leading cause of mortality among women in developing regions, where barriers to access and a shortage of specialists hinder the timely detection of precancerous lesions. Although methods such as colposcopy exist, their effectiveness relies heavily on the examiner's visual expertise, often leading to subjective diagnoses. In response to this reality, this study introduces Cérvix Tool, a mobile application designed to broaden access to AI-assisted screening. Unlike conventional systems that rely on external servers, Cérvix Tool integrates YOLO11, a single-stage deep learning architecture optimized to run directly on the mobile device, enabling operation without an internet connection. The system segments and analyzes cervical lesions in real-time, translating visual findings into a risk estimation based on the clinical Swede Score scale. The results validate the tool's ability to identify high-grade lesions with high sensitivity and minimal latency, providing an accessible and efficient triage solution for low-resource environments.