AI-CAD enhances pulmonary TB detection and yield in active case finding

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Public Health

Background:

• India faces the world's highest tuberculosis (TB) burden, accounting for 27% of global incidence.
• A community-based active case-finding TB screening program was conducted in Tamil Nadu, India.

Purpose of the Study:

• To evaluate the performance of an AI-assisted computer-aided detection (AI-CAD) solution for TB screening.
• To compare AI-assisted screening with conventional screening methods.

Main Methods:

• Digital X-ray machines in mobile diagnostic units were used for community-based TB screening.
• AI-CAD performance was assessed using AUROC, sensitivity, and specificity.
• Comparative analysis was performed against conventional screening data from five districts.

Main Results:

• AI-CAD achieved a sensitivity of 0.93 and specificity of 0.83, exceeding WHO targets.
• AI interpretation was strongly linked to positive TB diagnoses (OR: 58.95, P < 0.0001).
• AI-assisted screening resulted in a 2.09-fold increase in TB diagnoses and a 2.86-fold higher sputum positivity rate.

Conclusions:

• The AI-CAD solution met and surpassed the WHO's minimum target product profile for TB detection.
• The enhanced sputum-positive yield highlights AI-CAD's potential for large-scale TB screening programs.