As part of its bid to break into the biotech space — specifically, the rapidly expanding market for biological image analysis — Seattle-area SVision recently won $100,000 from the NIH to help it develop its quantitative cellular analysis microscopy software.
The Small Business Innovation Research grant, awarded in February, was SVision's second, and was administered via the National Institute of General Medical Sciences. The funds will go specifically toward building out cell motility analysis capabilities on the software, a company official said this week.
The company won its first $100,000 SBIR grant in August through the National Institute of Mental Health — a more general-purpose award to help it develop and sell the base quantitative microscopy software, called SVCell. A component of this software will be used for the cell motility-analysis product.
The biological imaging space is relatively uncharted territory for SVision, which has made its name since 1999 in quantitative image analysis for industrial applications, such as alignment, defect inspection, and classification in the semiconductor industry, Samuel Alworth, SVision's business and product manager, told CBA News this week.
According to Alworth, the company derives the bulk of its revenues, which are undisclosed, from contract work in this area. Since approximately 2002, however, SVision has been contemplating marketing its own software product, and identified biological image analysis as a rapidly expanding niche market.
"Anybody can create a recognition algorithm by simply drawing."
Part of the drive to biological applications came from the background of several of the company's employees.
"On the technical side we have a lot of experience with biological imaging," Alworth said. "Our team isn't new to this area." SVision's founder and president, James Lee, was previously the technical founder and CTO of a company called NeoPath, which created the first FDA-approved fully automated cytological cancer screening system, Alworth said. "Our principal engineer and probably three-quarters of the company were involved in NeoPath," Alworth said.
SVision is one of many companies in the last few years to begin competing in the biological image-analysis software space. Companies such as Definiens, Vala Sciences, BioImagene, and ImageInterpret have all recently released image-analysis products, while non-profit entities such as the Whitehead Institute have developed open-source image-analysis software.
SVision hopes to differentiate itself in a few major areas, however. First, most of the aforementioned companies entered into the space to ride the swelling wave of high-content cellular analysis and automated microscopy. SVision, for the time being, will aim to market its products to the basic microscopy research community.
Alworth said that "it's a little too early to say" whether SVision would eventually pursue HCS. "At this point, I can say that we're definitely intended to be compatible with laboratory microscopes," he said.
Second, SVCell will feature a unique approach to image analysis that greatly reduces the programming expertise required by the user. It's called soft-maxing technology, and if a user can draw an outline, basically he can use the software, Alworth said.
"Anybody can create a recognition algorithm by simply drawing," he said. "This is a type of computer learning. What we need from the user is a definition of what they want to identify, or a pattern to enhance; what they don't want, or a pattern to suppress; and then a background.
"They kind of draw these regions in a training image — and this can all be done on more than one image — and then we create enhanced images to which you can just apply a simple threshold," he added. "That's the basic technology."
Lastly, SVision purports that its software is intended to work across most major microscopy techniques used in biological research.
"We're saying any microscopy image," Alworth said. "We've taken in everything from [differential interference contrast] to phase [contrast] to electron microscope images — of course, fluorescence microscopy is still the bulk of the work."
As a case in point, the new SBIR grant will help the company develop a cell motility assay for the SVCell software for highly automated kinetic recognition of individual cells in time-lapse, phase-contrast images. This feature would find application in many areas, as cell motility is a key process in embryonic development, immune response, wound healing, angiogenesis, tissue engineering, and monitoring cell health in cell culture, among others.
SVision's first SBIR grant focused more on the basic capabilities of the software; specifically, the accurate quantification of subcellular functions. One specific example of this is the analysis of synaptic transmission in spines and synapses — one of the reasons NIMH serves as the grant administrator.
"We're looking at very small objects with this application, in particular something like a vesicle recycling assay of axon terminals labeled with dye," Alworth said. When looking at such structures, one microscope image can contain several hundred objects like these axon terminals, Alworth said. Typical analysis programs will use what is called a binary mask.
"Sources of variation — it could be computer error, biological variation, camera problems — if it introduces a whole-pixel error in an object that's, say, nine pixels, then all of a sudden you've got 10-percent error in your segmentation.
"So we're looking at using software representations called confidence masks," he added. "As opposed to binary masks, these have 256 levels, so you can give a confidence value to each pixel. Rather than one-zero, its one to 256 as a confidence value. The question we have is whether this can improve the results and the accuracy."
SVision does not yet have a product ready for launch, but Alworth said the company has just entered into its "early adopter" phase. He added that the company hopes to have a product on the market in the second half of this year. In the meanwhile, SVision will work to complete the Phase I portions of its SBIR grants in pursuit of the additional funding promised with a Phase II award.
Alworth added that the company also has the luxury of eschewing VC funding for the foreseeable future, as its machine inspection contract work is enough to support the company.
— Ben Butkus ([email protected])