High throughput demands fast diagnostic turnaround.
To properly provide a diagnosis of a stained sputum slide requires a microscopist approximately 7-10 minutes for each Auramine slide and 10-15 minutes for each ZN slide. They must manually refocus each time they move the slide to a new location, which contributes significantly to the length of time to process each slide.
To be successful we must not only focus the camera automatically, but we must also focus quickly.
Our automation software technology, through constant research and revision, performs the focus plane process in approximately one minute per slide. This autofocus capability allows the optimal z-position to be determined at various points on the slide before capturing the digital FOV images. These coordinates created for the system provide an instant focus reference when scanning the slide. Once the focus function is complete an additional one minute is required to capture the minimum 100 FOVs. Processing with this level of speed allows our system to automatically handle 200 slides in approximately 6-7 hours, which fits nicely within the normal working hours of the laboratory. Since the system requires no human interaction beyond initiation, it is possible to have an additional 200 slides ready for processing that occurs overnight with results available when the laboratory reopens in the morning.
Now that the technical team has developed several breakthroughs in auto-focusing without sacrificing high throughput capacity, attention turned to the performance of the detection algorithms as applied to the images being captured. In the universe of worldwide TB laboratories, TBDx™ must be sufficiently robust to efficiently function across a wide range of slide preparation extremes. While there are WHO endorsed protocols for the preparation of sputum smears, including specific instructions pertaining to each of the staining methods, individual country protocols vary. Some use thinning agents, others don’t. Some centrifuge specimens, some don’t. Some manually apply the stain, others use automated staining systems. The stains themselves vary widely due to several factors: (i) mix of chemicals, (ii) manufacturers of each chemical, (iii) imprecision of the volume mix, and (iv) the attentiveness of the individual preparing the stain.
While the stain and the staining process create variability, the more significant variable is the sputum application process. Sputum smears may be represented on the slide as very thick (creating focal plane issues) while others can turn out more thin and runny (creating contrast issues). Smear sizes can vary widely from very small to meandering all over the slide. Smear location changes with each slide prepared, so auto-focusing must first locate the boundaries of the smear to determine the starting point for FOV image capture. Detection algorithms that do not compensate for the wide range of smear variability cannot be successful. Our TBDx™ process normalizes the widest range of smear and staining variability to maximize the quality of the images captured – producing the highest achievable performance of our detection algorithms.
As you can imagine, this was no small task!