Artificial intelligence as a proficient tool in detecting pulmonary tuberculosis in massive population screening...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Public Health

Background:

• Tuberculosis (TB) remains a significant global health challenge, particularly in low-resource settings.
• Mobile screening programs are crucial for reaching underserved populations.
• Accurate and efficient diagnostic tools are needed to enhance TB detection rates.

Purpose of the Study:

• To assess the performance of Genki, an AI-based computer-aided detection (CADe) software, for TB detection using chest radiography.
• To evaluate Genki's utility within a mobile TB screening program in Chennai, India.

Main Methods:

• Genki software was deployed in four mobile diagnostic units for TB screening.
• Chest X-ray scans were analyzed by Genki, classifying them as 'TB suggestive' or 'TB not suggestive'.
• Confirmatory diagnostic tests were performed on patients with 'TB suggestive' results.

Main Results:

• Over 25,000 patients were screened between January and December 2022.
• Genki achieved a sensitivity of 98%, specificity of 96.9%, and accuracy of 96.9% in TB detection.
• High performance metrics were consistent across different sexes and age groups.

Conclusions:

• Genki demonstrates significant value as a TB screening tool in remote areas.
• The CADe approach offers a cost-effective solution for TB screening in high-burden, low-resource settings.
• AI-powered tools can effectively reduce workload and improve TB detection efficiency.