SVision, an image-analysis firm founded in 1999 to provide contract R&D services across a broad range of industries, has targeted the cell-based assay market for its first off-the-shelf software product.
SVCell, which the company is launching next week at the Society for Neuroscience conference in Atlanta, offers a “teach-by-example” interface that enables researchers to train the software to recognize features of interest in microscopy experiments.
Sam Alworth, business and product manager at SVision, told Cell-Based Assay News that the user-friendly interface should differentiate SVCell from other products on the market — from general-purpose image-analysis toolkits to more specialized cell analysis packages.
“What’s different about SVCell is that you can teach the software to do what you want, so it’s quite a different paradigm from the existing software that’s out there,” he said.
SVCell provides image segmentation, measurement, and classification capabilities, but Alworth said that the key component is the segmentation piece, which includes a drawing tool that allows users to train the software by specifying areas that they want to enhance, regions of background, and regions that are not of interest.
Using an example of a gap junction image with non-specific staining in the nucleus, Alworth said, “You want to get rid of that nuclear staining so you can just extract the cell boundaries.” With SVCell, “you can just circle the cell boundary and say, ‘I want this type of pattern.’ You can circle the nucleus and say, ‘I don’t want this kind of pattern,’ and you can give the software the background pattern information.”
The software analyzes each pixel in the image to determine which segment it belongs in, Alworth said. “It generates an enhancement image, and then the enhancement image can be processed in a normal fashion with a threshold and followed by binary operations.”
Current image analysis packages are developed for specific applications and don’t offer much flexibility, Alworth said — especially for biologists who are not comfortable programming software. “With our approach, when you draw, you’re actually creating an algorithm. The algorithm is behind the scenes, and it’s dependent on your specification when you draw on the image.”
The company calls these user-defined algorithms “recipes.” SVCell ships with a certain amount of predetermined recipes, “so you’re not always drawing,” Alworth said.
One benefit of the approach is that recipes can be updated very quickly based on new data. “You could have an image set that was taken in one facility by one operator and another set that was taken by another operator, and there could be some subtle differences that could break assumptions that would be built into a typical algorithm,” Alworth said. With SVCell, “you can easily update your recipe by training on images from the new set.”
Alworth said that the speed and accuracy of the software are comparable to other software packages, but did not provide specific details.
The company is first targeting the basic research market for the software, and has launched a trial program called SVCell Easy Deployment that will give academic labs free access to the software and free support for up to a year.
“In basic research there are so many different kinds of applications and that’s why we think it’s a great place to start,” Alworth said. “And we want to make sure we can satisfy.”
SVision’s contract research business is profitable, so the firm’s success doesn’t hinge on sales of the software, Alworth noted. “The SVCell launch isn’t all about making a lot of money right now. It really is about making a very good product as kind of a starting point and seeing where it goes from there.”
“What’s different about SVCell is that you can teach the software to do what you want, so it’s quite a different paradigm from the existing software that’s out there.”
While the company does plan to scale up the software to higher throughput eventually, Alworth said that SVision wants to ensure that the software meets the needs of researchers first. “We want to make sure that we can give people a versatile and accurate and satisfactory outcome for quite a broad range of applications,” he said. “So we’re not so much setting this up to analyze well plates, though that’s something that is on our roadmap.”
Also on the roadmap is another software package for live cell assays with capability for time lapse tracking and kinetics analysis. Development for that product is funded by three separate Small Business Innovation Research grants from the National Institutes of Health totaling $300,000.
Alworth said that SVision expects to launch the live cell analysis package by the third quarter of 2007.