March 19, 2014
APVS and FIND Collaboration Targets Automated TB Detection in Developing Countries
Collaboration to evaluate TBDx™ computer-vision software platform
Leesburg, VA (19 March 2014) – APVS (Applied Visual Sciences, Inc.), a software developer of next-generation image analysis applications and FIND (Foundation for Innovative New Diagnostics), today announced a feasibility project that will evaluate the performance characteristics of the TBDx™ automated slide management and tuberculosis detection platform in a field setting of FIND’s choice, and operating from a study protocol cooperatively developed with APVS. Under similar evaluation protocols, TBDx™ has demonstrated significant improvements in sensitivity without negatively impacting specificity by fully automating the sputum microscopy process and eliminating reader subjectivity.
Over the past four years APVS has been developing and training software to detect mycobacterium tuberculosis (MTB) in digital images acquired from Auramine-O stained sputum smears, using fluorescent microscopy. During this time, and with the assistance of the Aurum Institute for Health Research and the Center for Tuberculosis in Johannesburg, South Africa, the detection algorithms have continuously improved. In the latest TBDx™ performance study conducted at the Center, and presented during the Union World Conference in Paris this past fall, the software achieved sensitivity ranging from 74% to 80%, and specificity ranging from 80% – 96%. One study element highlighted the technology as a triage diagnostic, prior to a molecular testing on the most likely TB positive cases, and it produced substantial reductions in laboratory diagnostic costs.
The fully-automated platform includes a fluorescent microscope and camera, a movable slide stage, with an attached computer that receives high-quality digital images acquired from the camera. The computer operates detection algorithms that segment, evaluate, and classify objects of interest in these images. The platform is able to integrate an optional 200-slide automated slide loader for high volume settings. The application is user-definable and can capture 300 or more digital fields-of-view. The application can classify a case as positive (with appropriate WHO designation) or negative in 5 minutes or less.
“Participating with FIND in this important evaluation represents our mutual commitments to introduce to the Public Health Sector flexible and affordable diagnostics that will apply the brakes to this TB epidemic,” noted APVS President, Bill Donovan. “Smear microscopy presents unique advantages for our Signature Mapping™ digital analysis technology. Operating costs are relatively low. There is an extensive body of scientific evidence surrounding it. The laboratory infrastructure is in place. Our goal has been to leverage these in-place advantages using image analysis science to improve its’ performance. We believe the platform can be an affordable ‘triage’ diagnostic technology working in tandem with molecular testing in a layered diagnostic protocol.”
Donovan went on to note, “computer-vision and image analysis science will reimagine smear microscopy. Detection algorithms never tire, they work while others sleep, they don’t require training, and they produce consistent results. The technology can analyze a stained sputum specimen faster, with greater accuracy and consistency, wherever a microscope operates today. This technology can have a profound impact on this TB epidemic. The relevance of this announcement is especially meaningful as we recognize World TB Day on Monday, March 24th“
The evaluation will begin as soon as the study details have been finalized. Given the successful outcome of the FIND Study, the next step would be to evaluate and develop a joint plan to bring the TBDx™ system to the Public Health Sector of Developing Countries.
Terms and conditions of such a joint plan would be the subject of a separate mutually agreed Definitive Agreement.
Applied Visual Sciences, Inc.
Sean W. Kennedy
Office: 703-539-6190, Ext. 704