Tuberculosis is currently the world’s leading cause of adult death from a single infectious disease. When appropriately diagnosed and treated, TB is largely curable. Yet, in the most recent annual global statistics, 8.8 million people became ill with the disease and an estimated 3.1 million with active TB were misdiagnosed (laboratory failure to detect) and untreated. The inadequate ability to rapidly and accurately diagnose active TB in developing countries remains a major obstacle in the global control of the disease.
The most commonly used technique to diagnose TB is smear microscopy. Smear microscopy is a laboratory process whereby sputum specimens are affixed to a slide via the application of direct heat, then stained with either a fluorescent chemical (Auramine O) or a Ziehl-Neelsen process, which is an acid-fast stain. The stained slides are placed under either a fluorescent microscope or a bright field microscope to be read by a laboratory technician. The technician reads approximately 100 fields-of-view (FOV) for each slide, in accordance with World Health Organization protocols, in order to reach a diagnostic finding. On any given slide there are approximately 5,000 potential FOVs.
Benefits of smear microscopy:
- The process has been around for over a hundred years and is widely understood
- It is the primary diagnostic methodology across the developing world
- It is highly effective in diagnosing the most infectious cases
- It is the lowest cost method of diagnosis – always a primary factor for a developing country
- It can be performed in the most basic laboratory setting
Drawbacks of smear microscopy:
- Very low sensitivity (sensitivity is the ability to identify a positive case when presented) that fuels the undeterred transmission of the disease (undetected positive patients infect approximately 25 additional people)
- Process is very labor intensive
- Process requires a highly trained staff
- Process is very tedious – 25 to 60 slides per day per technician
- Is only as good as the human technician
- Lacks diagnostic consistency due to levels of training, experience and other human factors
It was the inherent challenges of diagnosing a ravaging disease in the developing world that brought Dr. Dave Clark, Deputy Chief Executive Officer for The Aurum Institute, to APVS. Dr. Clark had previously visited the APVS exhibit booth at the Radiology Society North America conference in Chicago and was impressed with the early work we had done in breast cancer detection. He saw a bridge from radiology to pathology. It was his vision that foresaw the potential benefits of combining computer-vision detection capabilities to smear microscopy. In the initial product design, developed in cooperation with Aurum and the South African National Health Laboratory Services, TBDx was intended to be retrofitted to an existing microscope and to provide an image analysis capability that would require the laboratory technician to focus the camera and manually capture the digital slide images for analysis. As is widely known, TBDx has expanded to be much more than the original vision could have imagined. But more on that in future blogs.
We always knew that computer-vision technology would improve diagnostic sensitivity and consistency by removing the human from the image analysis and decision process. We also knew that the application would fit economically within the financial framework of these developing countries. What we didn’t know when we started the product development was that we could eliminate the human from the process entirely, automate the capture of high-quality digital images and perform the complete analysis in 2-3 minutes per slide, providing an 8-hour shift capacity of 200 slides without human intervention.
Lastly, our analysis of the global marketplace indicated an extremely large and underserved market. Globally, on an annual basis, there are approximately 100 million smear tests performed, excluding smear tests performed in Australia, Europe, Japan, New Zealand, and North America. These numbers represent those individuals who presented to a medical facility and were clinically diagnosed with TB symptoms. In studies conducted in India, and which probably holds true for other countries, approximately 30% of all active TB patients present for diagnosis at a medical facility. The remaining 70% represent the vulnerable and marginalized segments of society – migrant workers, slum dwellers and the transient poor. To break the transmission of the disease will require that a greater percentage of the population be screened for the disease through an outreach mass population screening program. Such a program will greatly expand the economic potential of our fully automated, computer-vision TBDx system.