Using Ai for Screening Tuberculosis

[Kevin Zam]

Development and evaluation of an artificial intelligence (AI) -assisted chest x-ray diagnostic system for detecting, diagnosing, and monitoring tuberculosis (Research Abstract)

Objectives: To develop an artificial intelligence (AI)-assisted chest x-ray diagnostic system for the detection, differential diagnosis, and follow-up of tuberculosis (TB), and prove its usefulness.
Methods: This is a retrospective study. In-house developed AI-assisted chest x-ray diagnostic system was used to identify and diagnose lung abnormalities in participants’ chest x-rays and to compare imaging findings from two x-rays. First, 100 chest radiographs were reviewed including TB cases (N ¼ 43) with positive sputum test confirmation and non-TB cases (N ¼ 57) for initial diagnosis and differential diagnosis. Next, 45 pairs of TB cases from the identical patients were reviewed for follow-up. The AI system diagnosed TB and graded the comparison images into three categories (improved, stable, or worsening). The performance was evaluated by four expert radiologists or pulmonary medicine specialists.
Results: The AI system demonstrated an exceptional sensitivity of 100 %, successfully identifying all 43 TB cases. Nevertheless, it is also susceptible to misclassify other diseases as TB, resulting in low specificity score of 66.7 %. The comparison function determined that expert physicians and AI-assisted chest x-ray diagnostic system were 58 % in exact agreement and 100 % in within one grade agreement.
Conclusions: The AI system successfully detected all TB patients identified in this study and demonstrated a reasonable comparison function. Therefore, our AI assisted chest x-ray diagnostic system is feasible and practical for TB screening.
This is an open access article retrieved from HERE.

1. Brief description of the eHealth project
The research project focuses on the development and evaluation of an AI-assisted chest x-ray diagnostic system for detecting, diagnosing, and monitoring tuberculosis (TB).
It was created using large datasets of chest x-rays from international sources and local public health centers in Thailand.
The AI system uses deep learning algorithms (UNet with EfficientNet backbone) to:
-Detect TB-related abnormalities (cavitation, infiltration, nodules).
-Estimate the likelihood of TB (none, low, medium, high).
-Compare chest x-rays over time to monitor treatment outcomes (improved, stable, worsening).
The system was tested against radiologists’ interpretations and demonstrated 100% sensitivity in detecting TB cases, though with lower specificity for non-TB diseases.

2. Out of 3 eHealth Domains (Health in our hands, Interacting for health and Data enabling health), this research project fit into Data enabling health because of the fact that using AI which relies on large-scale imaging datasets to produce structured outputs (likelihood scores, progression categories), which can strengthen TB surveillance systems, public health reporting, and research.

3. The project adds value in several ways:
Cost efficiency
Reduces dependency on specialized radiologists in low-resource settings.
Minimizes costs related to delayed diagnosis and advanced TB treatment by enabling earlier detection.

Customer (patient) satisfaction
Shorter waiting times for results, reducing anxiety.
Increased accessibility for patients in remote or underserved areas through mobile x-ray and AI systems.

Improved outcomes
Early and accurate TB detection (100% sensitivity in the study).
Enhanced treatment monitoring by reliably categorizing patient progress.
Potential contribution to WHO’s End TB Strategy by improving case detection and follow-up care.

Overall, the AI-assisted chest x-ray system is a feasible, practical, and scalable innovation that can strengthen TB screening, triage, and monitoring in both urban hospitals and rural clinics

[Myo Oo]

Thanks a lot for sharing. This project is very interesting because it makes TB screening more accessible and cost-efficient, especially in low-resource areas. It would be better if the system could balance high sensitivity with better accuracy.

[Than Htike Aung]

Thank you for sharing this project. This kind of projects opens new opportunities for massive screening of general population without heavy resources. This will greatly improve the health outcomes related to TB especially in our country like Myanmar. Without labour intensive screening work and rapid processing time will not only reduce the cost but also improve customer satisfaction.

[Saranath]

Thanks for sharing the interesting project. I was wondering if this tool has been implemented in the real clinical settings. For the improvement on your presentation, 1) you took almost 10 minutes for presentation (the provided time is 3-5 minutes); 2) you may improve on visual aid, making slides more attractive.

[Author]

Posts