Tuberculosis (TB) is the second deadliest infectious disease worldwide. Current TB diagnostics utilize sputum samples, which are difficult to obtain, and sample processing is time-consuming and difficult. This study developed an integrated diagnostic platform for the rapid visual detection of Mycobacterium tuberculosis (Mtb) in breath samples at the point-of-care (POC), especially in resource-limited settings. The less pathogenic Mycobacterium smegmatis containing same gene fragment of Mtb served as the model bacterium. A novel respirator was designed to collect airborne mycobacteria in breath samples, with an efficiency of 38.7-61.5 % (10