Automated chest radiography and mass systematic screening for tuberculosis

Area of Science:

• Public Health
• Infectious Disease Epidemiology
• Medical Imaging Technology

Background:

• Large-scale implementation of automated chest radiography (ACR) with computer-aided detection software (CAD4TB) for tuberculosis (TB) screening in Karachi, Pakistan.
• Limited real-world data on ACR/CAD4TB effectiveness in mass screening settings, despite established evidence for pre-screening presumptive TB patients.

Purpose of the Study:

• To evaluate the performance of ACR with CAD4TB as a pre-screening tool in a high-burden setting.
• To assess the association between CAD4TB scores and Xpert MTB/RIF positivity during mass screening.

Main Methods:

• Mobile digital X-ray units used for screening individuals aged ≥15 years in hospital and community settings.
• Chest X-rays analyzed by CAD4TB software; individuals with a score ≥70 were offered Xpert MTB/RIF testing.
• Data collected from January 1 to June 30, 2018, using a custom data collection tool.

Main Results:

• Over 127,000 individuals screened, with 9.1% having a CAD4TB score ≥70.
• Xpert positivity increased significantly with higher CAD4TB scores, from 0.7% (<50) to 23.5% (>90).
• A strong linear association between CAD4TB scores and Xpert positivity was observed in both community and hospital settings.

Conclusions:

• ACR with CAD4TB demonstrates strong performance as a large-scale pre-screening tool in high-burden TB settings.
• The findings support the utility of ACR-based pre-screening in real-world public health initiatives for TB detection.