Raman spectroscopy and machine learning-based optical probe for tuberculosis diagnosis via sputum

Area of Science:

• Biomedical Engineering
• Spectroscopy
• Machine Learning

Background:

• Tuberculosis (TB) is a significant global health threat, causing millions of deaths annually.
• Current diagnostic methods, like sputum culture, are time-consuming, delaying critical treatment initiation.
• There is an urgent need for rapid, accurate, and accessible TB diagnostic tools.

Purpose of the Study:

• To develop and validate a portable, non-invasive optical sensor for rapid tuberculosis detection.
• To utilize Raman spectroscopy combined with machine learning for enhanced diagnostic performance.
• To assess the sensor's accuracy in identifying TB patients, including those undergoing medication.

Main Methods:

• Development of a non-invasive optical sensor utilizing Raman spectroscopy on sputum supernatant.
• Application of Principal Component Analysis (PCA), a machine learning algorithm, to analyze spectral data.
• Clinical testing of the sensor on 112 potential tuberculosis patients.

Main Results:

• The developed sensor achieved 100% accuracy for true-positive TB detection.
• A true-negative accuracy of 93.4% was recorded, demonstrating high specificity.
• The sensor successfully identified patients undergoing tuberculosis medication.

Conclusions:

• The developed optical sensor provides a rapid, accurate, and non-invasive method for TB diagnosis.
• The integration of Raman spectroscopy and machine learning significantly enhances diagnostic capabilities.
• This technology holds promise for a viable and rapid TB diagnostic platform, improving global health outcomes.