Implementing a chest X-ray artificial intelligence tool to enhance tuberculosis screening in India: Lessons learned

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Public Health

Background:

• Artificial Intelligence (AI) based chest X-ray (CXR) screening for tuberculosis (TB) is gaining traction.
• Deployment of AI tools faces significant real-life barriers, including software installation, workflow integration, and network connectivity.
• Resource-limited settings present unique challenges for implementing advanced diagnostic technologies.

Purpose of the Study:

• To evaluate the operational feasibility and accessibility of an AI software device (qXR) for TB screening in a resource-limited setting.
• To identify and address real-life implementation challenges for AI-based TB screening.
• To assess the impact of AI on TB case finding in an active case-finding program.

Main Methods:

• Implementation of the qXR AI software in eight private CXR laboratories in Nagpur, India.
• Screening of 10,481 presumptive TB individuals identified through informal providers and clinical history.
• Comparison of TB case detection by radiologists and AI interpretation of CXR images.

Main Results:

• AI screening identified an additional 15.8% of TB cases that were not flagged by radiologists.
• The program screened a substantial number of individuals, demonstrating the potential for large-scale application.
• Key lessons were learned regarding operational feasibility and strategies to overcome implementation barriers.

Conclusions:

• AI-based CXR screening, exemplified by qXR, can significantly enhance TB case finding in resource-limited environments.
• Overcoming implementation challenges is crucial for the successful and seamless deployment of AI tools in healthcare.
• Further integration of AI into diagnostic workflows is essential for improving TB control efforts globally.