Bangladesh – Improving TB case detection in remote areas with Delft Light & CAD4TB
The National TB Control Programme (NTP) introduced ultra-portable chest X-rays with AI to improve TB case detection in remote areas and facilitate the roll-out of TB Preventive Therapy (TPT).
Screening in remote districts
Six Delft Light units were deployed across six districts in the Rajshahi Division.
Two groups of participants
Participants were divided into two groups: household contacts and TB presumptive identified by clinicians.
AI-guided presumptive identification
A CAD4TB cutoff score of 50 was used alongside symptom screening.
Confirmatory testing
All TB presumptive identified by X-ray and symptom screening were referred for bacteriological confirmatory tests and clinical evaluation.
From January to March 2023, 5,137 TB presumptive were screened by Delft Light. Of these,
- 7.22% (43) were bacteriologically confirmed TB (B+), and 22.92% (119) were clinically diagnosed with TB (CD) while they had Bac-negative.
- 85.75% (4,405) had CAD4TB score <50, where 0.43% (15/3,484) were B+ and 0.79% (21/2,633) were CD.
- A significant difference was observed in TB case detection using AI software among CAD4TB scores ≥50 and <50.
- When comparing CAD4TB cutoff scores 50 and 40, a score ≥40 increases the number of TB cases diagnosed, which means fewer cases are missed.
- Out of the over 32,000 data, ≥40 would miss less than 4% of people with TB.
- Ultra-portable X-rays with CAD4TB can be a game-changer in detecting TB cases early, including asymptomatic cases.
- They also help save cartridges and ensure equitable access to quality care.
- NTP can scale up this technology to diagnose missing TB cases in remote locations and for nationwide TPT roll-out.
REFERENCE: Rahman, M. (2023, November 15-18). Ultra-portable X-ray with artificial intelligence for TB screening and facilitating TB preventive therapy in Bangladesh [Conference session]. The Union World Conference on Lung Health 2023, Paris, France.