DecXpert

DecXpert detects active tuberculosis from chest X-rays using deep neural networks to identify subtle radiological abnormalities for early diagnosis in settings with limited skilled radiological interpretation.

Key Features:

• Deep Neural Network-Based Detection: DecXpert employs deep neural networks for computer-aided detection of tuberculosis-related abnormalities in chest X-rays.
• Demographic Integration: Incorporation of patient demographics improves diagnostic performance, achieving an area under the curve of 0.91 (95% CI 0.88-0.94).
• Screening Deployment: Configured for deployment as a screening tool in resource-limited settings where skilled radiological interpretation is scarce.

Scientific Applications:

• Active Tuberculosis Screening: DecXpert enables early identification of individuals with active TB disease in primary health care and tertiary hospital settings.
• Resource-Constrained Diagnostics: DecXpert facilitates TB management in resource-limited settings by providing radiological diagnosis without requiring skilled human interpreters.
• Validation Against Molecular Gold Standards: DecXpert performance can be validated against GeneXpert MTB/RIF for assessing diagnostic accuracy in target populations.

Methodology:

DecXpert version 1.4 was validated in a multicentre study analyzing data from 4363 individuals across 12 primary health care centers and one tertiary hospital in North India, with performance evaluated against GeneXpert MTB/RIF.