Usage of artificial intelligence tools in community-level X-ray triaging for tuberculosis in Chennai, Tamil Nadu

Abstract

Background: The end tuberculosis (TB) strategy emphasises early and correct diagnosis of TB. Chest X-ray (CXR) is an essential tool for triaging and screening TB and confirming the diagnosis in fewer situations. Greater Chennai Corporation (GCC) implemented Mobile Diagnostic Units (MDUs) retrofitted with X-rays with artificial intelligence (AI). Objectives: The study's objectives were to determine the X-ray triaging performance in MDU vans using AI tools in GCC, Tamil Nadu. Materials and Methods: AI is to increase access to quality TB screening diagnostics in high-risk locations. Genki AI-powered Public Health Screening Solution from Deeptek used for TB triaging after uploading CXR images from MDU. X-ray has been uploaded in AI software once taken, and the results were available immediately after uploading. The radiologist reports helped to take further courses of action. Results: A total of 79,462 CXR was taken from April 2019 to April 2022 from 7 MDU vans. Amongst 3.4% were identified as suggestive of TB, 1.4% old TB, 0.89% COVID (from 2020) and 7.2% other chest abnormalities. The sensitivity of CXR-AI was 0.98 (95% confidence interval [CI]: 0.97, 0.98), and the specificity was 0.96 (95% CI: 0.96, 0.97). Conclusion: AI helps in faster triage for further public health action and eliminates the challenges of the availability of functional X-rays, interpretation and reporting.

Keywords

• artificial intelligence
• chest x-ray
• machine learning
• national tb elimination programme
• public health programme
• triage
• tuberculosis