Automated identification of tubercle bacilli in sputum. A preliminary investigation

Abstract

Objective: To use an automated method to detect tubercle bacilli in sputum specimens.

Study design: Using fluorescence microscopy, tubercle bacilli were identified on auramine-stained sputum specimens. Images were then captured with a digital camera and enhanced through imaging processing techniques. The bacilli were recognized using neural network classifiers.

Results: This preliminary investigation demonstrated a sensitivity of 94.1% for the identification of individual bacilli. As there are usually fairly numerous tubercle bacilli in the sputum of patients with active pulmonary tuberculosis, the overall diagnostic accuracy of sputum smear-positive patients can be expected to be very high.

Conclusion: Potential benefits of automated screening for TB bacilli are: rapid, accurate, inexpensive diagnosis; the ability to screen larger numbers of people; increased resources to monitor patients; and reduction in health risks to staff.