Case Study

Nigeria – Community Active Case finding with Delft Light & CAD4TB

Delft Light closeup

In Kano State, AI-supported chest X-ray screening using CAD4TB demonstrated higher efficiency and effectiveness than the WHO four-symptom screen for community TB case finding. The approach improved TB detection, reduced the number needed to test, and enabled more efficient use of diagnostic resources.

Community active TB case finding

Community-based active case finding was implemented in Kano State to improve early TB detection.

Using ultra-PDX with CAD4TB

Chest X-ray screening supported by CAD4TB was used to identify individuals with presumptive TB.

Comparison with symptom screening

The performance of CAD4TB-supported X-ray screening was compared with the WHO four-symptom screen.

Confirmatory TB testing

Presumptive TB cases from both screening approaches were evaluated bacteriologically.

Result

Among those screened conducted for the study, with method CAD4TB-supported X-ray screening

9 43 people
identified as presumptive TB (CAD4TB score >60)
1 16 (12%) people
diagnosed with TB

and with method WHO four-symptom screening

4 ,272 people
identified as presumptive TB (CAD4TB score >60)
1 15 (3%) people
diagnosed with TB
  • The difference in presumptive TB and case yield from both arms was statistically significant.
  • The number needed to test (NNT) was 5 using the CXR screen and 39 using W4SS.
Delft Light closeup
Conclusion
  • CXR screening with CAD4TB performs better than W4SS correlating with bacteriologic positive TB results.
  • A lowered NNT bears the advantage of cost saving with more efficient use of sputum cups and Xpert cartridges.
  • Bringing to scale AI-aided CXR screening would be a cost-efficient way for early TB detection in similar settings.

REFERENCE M. Bajehson et al. (2022, November 8-11). The use of artificial intelligence software aided chest Xray screening for community active case finding in Kano, Nigeria. The UNION World Conference on Lung Health 2022.