AI-assisted VIA Screening of Cervical Cancer

Cervical cancer is the fourth most common cancer among women globally, causing 90% of deaths in Low and Middle-Income Countries (LMICs). While easily curable if diagnosed and treated early, Nepal lacks accessibility to early-stage screenings. Visual Inspection Using Acetic Acid (VIA) is a simpler, inexpensive, and accessible test suitable for low-resource settings, but its effectiveness is subjective and dependent on human expertise, which is often unavailable in primary and community healthcare.

Our goal is to build AI- and machine learning-assisted VIA screening tools using mobile-captured photos to provide more consistent and accurate smartphone-based diagnostics, specifically for nurses and community health workers in low-resource settings.

The project establishes a robust smartphone-based AI-assisted system that does not require separate integrated devices. It includes a proposed protocol for quality image acquisition in resource-constrained settings, a dataset collected from 1,430 women during VIA performed by nurses in screening camps, a preprocessing pipeline, and training and evaluation of a deep-learning-based classification model aimed to identify (pre)cancerous lesions. The work demonstrates that readily available smartphones and a suitable protocol can capture cervix images with the required details for the VIA test, and the deep-learning-based classification model provides promising results to assist nurses in VIA screening, offering a direction for large-scale data collection and validation in resource-constrained settings.