Cancer is a growing global health concern responsible for approximately 10 million deaths annually, with low- and middle-income countries (LMICs) accounting for over 70% of cancer-related mortality. Nigeria, the most populous country in Africa, bears the highest incidence and mortality burden. Although Artificial Intelligence (AI) holds promise for improving early detection and treatment, its success depends on access to large, high-quality imaging datasets. Nigeria lacks a centralized repository for cancer imaging data to support AI development. This study proposes a framework for establishing a national cancer imaging repository to facilitate data-driven cancer research and AI model development in Nigeria, using twelve (12) tertiary healthcare institutions across the six geo-political zones as pilot implementation centres. It suggests a set of protocols to guide the standardization of the repository’s operations. A robust data ingestion and pre-processing pipeline was also designed to ensure that the repository maintains only high-quality images, adheres to ethical standards and supports interoperability with existing databases. Using a set of criteria, the study suggests twelve hospitals spread across the country’s six geopolitical zones to serve as the pilot data collection sites. The study also outlines the infrastructural requirements for the initial repository deployment and provides a roadmap for national scale-up in the future. The proposed architecture was benchmarked against The Cancer Imaging Archive (TCIA) to validate the proposed framework. The comparison assessed metadata standards, data governance, access protocols, and security features. The results show that the framework aligns with international best practices while addressing Nigeria’s infrastructural and regulatory constraints. The proposed repository framework provides a secure and scalable foundation for AI-powered cancer diagnosis and research in Nigeria. Its alignment with global standards enhances its potential to support national cancer control strategies and foster innovation in low-resource settings.
Adegoke-Elijah et al. (Fri,) studied this question.