Detecting drug-resistant tuberculosis in chest radiographs

Area of Science:

• Medical imaging
• Machine learning
• Infectious diseases

Background:

• Tuberculosis (TB) remains a global health crisis, with drug-resistant TB posing a significant challenge to control efforts.
• Timely diagnosis of drug-resistant TB is vital for effective treatment and preventing further transmission.

Purpose of the Study:

• To investigate the potential of using chest X-rays and machine learning for automated discrimination between drug-sensitive and drug-resistant tuberculosis.
• To establish a computational method for early identification of drug-resistant TB.

Main Methods:

• Utilized image analysis and machine learning techniques, specifically an artificial neural network.
• Extracted shape and texture features from chest X-rays for classification.
• Evaluated the model's performance using the area under the receiver operating characteristic curve (AUC).

Main Results:

• Achieved an AUC of up to 66% in discriminating between drug-sensitive and drug-resistant tuberculosis.
• The inclusion of follow-up X-rays did not significantly improve the discrimination performance.

Conclusions:

• Chest X-rays contain exploitable information for predicting drug-resistant tuberculosis likelihood.
• Further validation on a larger dataset is recommended to confirm these findings.