Applied Visual Sciences, Inc. (APVS.OB)
Over the past several months TBDx has been engaged in evaluations in several countries or has been involved in the publication of those results. Here is a quick update:
- The results from the TBDx evaluation in South Africa were published on June 15th in the peer-reviewed American Journal of Respiratory and Critical Care Medicine. The article is entitled, Performance of a Novel Algorithm Using Automated Digital Microscopy for Diagnosing Tuberculosis.
- Appearing in the same publication is an editorial entitled, Automated Digital Microscopy in New Tuberculosis Diagnostic Algorithms: Can it Boost Case Finding? Jose A. Caminero and Dr. Giovanni Battista Migliori authored the article. Dr. Caminero is affiliated with Dr. Negrin University Hospital of Gran Canaria in Las Palmas, Spain and with the Multi-Drug Resistant Tuberculosis Unit of the International Union Against Tuberculosis and Lung Disease (IUTLD). Dr. Migliori is affiliated with the World Health Organization in Tradate, Italy. The editorial, while complementary of the study and of the digital imaging technology, acknowledges that TBDx could be a “bridging” technology until such time as molecular tests are more affordable and more prevalent in the healthcare systems.
We agree with the authors that TBDx may represent a bridging technology. As technologists we see and actively participate in this disruptive process daily, and we realize that all technology eventually is displaced by newer, more innovative technology. In that regard, molecular testing also can be considered a bridging technology to the next wave of future innovation that we see emerging in laser technologies and photonics. The real challenge for all technology companies, ours included, is to accurately understand the pace of this anticipated replacement, something that is largely influenced by both price and performance considerations. Computer-vision software development and image analysis technology has a decided advantage in its cost structure, which greatly influences the price of the technology. It is easy to see the day when a TB test will be a fraction of a dollar.
Today the world of infectious diseases demands more affordable, more mobile, more responsive technologies that perform as triage to the more expensive molecular testing. In a scenario of limited global funding availability, the use of a computer-vision-based technology to weed out negative cases, with a high rate of predictive value, and focus the more expensive molecular testing on only those cases of potential disease will provide the highest sensitivity and specificity at the lowest overall cost of diagnosis, today. We see computer-vision technology working in cooperation with molecular tests rather than two technologies competing against each other.
As computer-vision technologies get better and better, with modest refinements in the software, the detection performance of TBDx will draw closer to that of molecular tests, putting downward pressures on the price of both technologies. Combined with the miniaturization of hardware platforms, such as mobile phones and tablets, our TBDx diagnostic platform will quickly evolve into an even less expensive miniature-imaging platform capable of diagnosing a dozen or more infectious diseases, each detection application employing computer-vision technology similar to that of IBM’s Watson.
- In the South Africa evaluation one important study element that required additional analysis was the economic impact of using TBDx as a triage tool. This analysis has been under the direction of Dr. David Dowdy of the Bloomberg School of Public Health at Johns Hopkins University. An economic model has been developed and the results are expected to be presented in first draft later this month.
- Data accumulation and the study evaluation in Nigeria has been completed. Dr. Luis Cuevas of the Liverpool School of Tropical Medicine, along with his colleagues, have drafted the outline of the initial version of the report, and it is anticipated that they will complete final drafting later this month.
- The FIND-sponsored evaluations in Peru and Vietnam have been completed. Though not all of the culture results are known yet, making the final results incomplete, we have received the preliminary findings and have observed the following:
- As discussed in previous blogs, both Peru and Vietnam evaluated TBDx using concentrated and direct smear slides. TBDx performance in Peru, irrespective of the smear preparation method, demonstrated similar performance to that of South Africa. In Vietnam the concentrated smear results were only marginally lower than South Africa, however the direct smear results are approximately 18 basis points lower. The lower sensitivity was not unexpected as the quality of smears has a direct impact on the quality of the images captured. Direct smears have more topological variation than concentrated smears and result in more objects being out of focus.
- Early analysis of the Vietnam cases missed by TBDx indicate that some cases classified incorrectly as TB Negative, in fact, included TB bacilli. However, the captured image was too blurred to facilitate the proper segmentation and classification of the case. Anticipating that we might encounter focus difficulties due to the nature of the direct smears, we had already developed an “image stacking” function to address this issue. The stacking function may have had minimal impact in Peru, but would most likely have produced improved performance in Vietnam. We elected not to deploy the stacking functionality so as to better assess the performance capabilities and limitations of TBDx.
- FIND hopes to complete a full analysis of both sites by the end of this month, at which time a go-forward strategy to achieve FDA acceptance will be determined.
All of these studies, and resulting data, represent a steady movement towards international classification of digital sputum microscopy as an acceptable alternative to routine smear microscopy and as a possible triage tool that either screens in or screens out the possibility of suspected TB.
Applied Visual Sciences, Inc. is a computer-vision software technology company publicly-traded on the OTC Bulletin Board under the trading symbol, APVS. The company has developed intelligent, next-generation imaging analytics and informatics technologies for the extraction, analysis, and detection of objects-of-interest within any digital image format (still or video).