Indonesia – Hospital-based TB screening using CAD4TB
In Indonesia, limited access to expert radiologists contributes to missed tuberculosis diagnoses, even within hospital settings. CAD4TB-assisted interpretation of chest X-rays may support clinical decision-making and improve TB detection.
National tertiary hospital
An analysis of adult chest radiographs from a national tertiary hospital was conducted.
Using CAD4TB
Images were interpreted using CAD4TB and independently reviewed by experienced radiologists, with bacteriological testing used as the reference standard.
Result
- At a defined CAD4TB threshold, the system achieved sensitivity above 80% with moderate specificity.
- Diagnostic performance was comparable to radiologist readings, with consistent specificity advantages across several clinical subgroups.
Conclusion
- CAD4TB can effectively support radiologists in hospital-based TB screening, even in high-burden settings with high case complexity.
- Its integration into clinical workflows may imporve efficiency and consistency of TB triage.
REFERENCE: Burhand E. et al. (2025). Pulmonary tuberculosis prediction using CAD4TB artificial intelligence among Indonesia hospital patients. medRxiv preprint.