At the end of this month we will be departing for Paris to participate in the 44th Union World Conference on Lung Health. From October 29th to November 3rd, the company will showcase TBDx™, its flagship TB diagnostic technology, to more than 3,000 attendees. Appointments are scheduled with representatives from global health policy organizations, leading universities and institutes dedicated to infectious diseases, industry journals and national TB programs from Africa, India, and Asia. On November 3rd Nazir Ismail, the Director of the Center for Tuberculosis in Johannesburg, South Africa will present the results of the TBDx™ technology evaluation completed this May.
During the Union/CDC Late-Breaker session the evaluation results and performance comparisons will introduce a new diagnostic to attendees, one that is an updated version of a familiar technology, consistently high performing and affordable, and a new entrant in the fight against the spread of tuberculosis. Whether operated as a stand-alone test, or in combination as a screening technology for a secondary molecular evaluation, computer vision technology will bring new advantages to TB detection. Attendees will learn that TBDx™, a re-imagined form of smear microscopy, offers a capability uniquely associated with image analysis software.
Today, culture is the “gold standard” by which each TB diagnostic technology is measured. Whether it is TBDx™, a molecular test, or routine smear microscopy used to evaluate a sample from a TB suspect, each of these is compared to culture results. Unfortunately, culture has limitations and is not 100% accurate, a characteristic shared by all TB diagnostics. In the case of culture or molecular testing, each requires thousands of bacilli to be present for the sample to grow in culture or for the bio-marker to be detected. Routine smear microscopy suffers from a similar problem, failing to detect the disease in “scanty” cases for the very same reason – too few bacilli on the slide – a leading reason why human vision performs at 50%-60% sensitivity.
Leveraging state-of-the-art computer-vision science, TBDx requires just one bacillus to be present in the image for detection. In the most recent TBDx™ evaluation the system acquired more than 300,000 digital images. The vast majority contained no TB bacilli. However, there were a few suspect cases that culture assessed as negative that TBDx™ and its computer vision technology classified as positive. Most often they were very low count (1 or 2) “scanty” cases. Nonetheless, culture is the final arbiter of TB Positive or Negative, regardless of the TBDx™ classification.
However, TBDx™ offers one unique and very important advantage over other technologies – the detection platform acquires digital evidence of its assessments that can be reviewed and evaluated further, if desired. Findings from these investigations can lead to new insights and, potentially, additional laboratory tests. We have prepared a document containing images illustrating this uniqueness. Unfortunately, clinical histories containing valuable patient information about previous disease, current treatment regimens, and HIV infection are not available currently. These would further inform the reviewer’s assessment of the image, certainly.
The official launch of TBDx™ is underway. If you are a blog subscriber expect an alert following the Late Breaker session when we report on the evaluation results and attendee reactions.