Artificial intelligence in medical imaging: Utilization, challenges, and practitioner perceptions in Rwanda

Artificial intelligence (AI) holds transformative potential for medical imaging in low-resource settings like Rwanda, where shortages of imaging professionals contribute to diagnostic delays. While global research has examined AI adoption in high-income countries, limited evidence exists for sub-Saharan Africa. This study aimed to assess AI utilization patterns, implementation challenges, and practitioner perceptions across Rwanda's healthcare system, providing critical insights for optimizing AI integration in resource-constrained environments.

A cross-sectional study was conducted from December 2024 to April 2025, surveying 107 medical imaging practitioners (74% response rate) across teaching, provincial, and district hospitals. Using a validated questionnaire, data were collected on AI utilization, implementation barriers, and practitioner attitudes. Descriptive statistics and chi-square tests analyzed patterns and associations.

Automated measurements were the most widely used AI application (69.2% of practitioners), while segmentation AI was the least adopted (7.7%). Utilization disparities were stark: 71.0% of teaching hospitals used AI versus 13.1% of district hospitals. Key barriers included high costs (27.1% of respondents), system integration challenges (26.2%), and training deficiencies (44.2% across all professional ranks, p > 0.05). Despite these challenges, 86% of practitioners viewed AI positively, with 37.4% strongly agreeing it improves diagnostic accuracy.

Rwanda's AI adoption in medical imaging shows promise but remains uneven, with teaching hospitals leading implementation. Systemic barriers, particularly costs, infrastructure limitations, and training gaps, must be addressed to ensure equitable expansion. The overwhelmingly positive practitioner attitudes indicate readiness for AI integration, suggesting Rwanda is poised for growth with targeted policy interventions. These findings provide a foundation for optimizing AI implementation in low-resource settings across sub-Saharan Africa. Future research should explore cost-effective scaling strategies and long-term impacts on diagnostic outcomes.