NOTE: Since the publication of this article, CBA News has learned that Petra Perner's presentation has been removed from the agenda of Cambridge Healthtech Institute’s High-Content Analysis conference. In addition, ImageInterpret is not officially affiliated with the HCA conference in any way, according to a CHI official.
This article has been updated from a previous version to clarify the fact that ImageInterpret is not demonstrating its software at the conference.
German firm ImageInterpret has become the latest software company to launch an image-analysis platform for the high-content screening market. The director of the German institute from which the company spun out two years ago will be showing off the program during Cambridge Healthtech Institute's upcoming High-Content Analysis conference in San Francisco.
ImageInterpret's high-content screening product, called Cell-Interpret, looks to make headway in an increasingly crowded market for independent image-analysis software. This market includes start-ups such as Vala Sciences, BioImagene, fellow German company Definiens, and various open-source academic projects, the most notable of which is at the Whitehead Institute for Biomedical Research.
Many high-content screening platform vendors, including Cellomics and Molecular Devices, also offer image-analysis software that can be used with competing instruments. However, these software packages are more often bundled with the imaging instrument from the same company, leaving many researchers little choice in image-analysis programs. Some researchers will use the bundled software but will look to other sources for specific add-on capabilities.
This, plus the fact that high-content screening has become more firmly entrenched in pharmaceutical research, has caused these start-ups to believe they can sell their image-analysis software to dissatisfied HCS instrument users.
"Our mission is that the system should give the final decision — so it's not just giving numbers, statistics, the sizes of objects, or other parameters, but the outcome is the decision a human expert would usually give."
Definiens, Vala, and BioImagene must still convince potential customers that their image-analysis software offers something better than what is bundled with HCS instrument platforms. Definiens, for one, claims that its method is unique because it is based on the type of cognitive object recognition originally used to interpret satellite imagery; Vala, meanwhile, has been touting the compatibility of its software with most every imaging platform on the market.
ImageInterpret's software appears to be closer to Definiens' in that the algorithms based on object recognition can be applied across a broad range of scientific images, from MRI scans to stained tissue slices to fluorescence microscope images.
ImageInterpret was spun out of the Institute of Computer Vision and Applied Computer Sciences (IBaL) in Leipzig two years ago. "We are developing [software] based on machine learning, data mining, and case-based reasoning," IBaL Director Petra Perner, told CBA News. "And we are developing flexible image-analysis and image-interpretation methods based on these techniques.
"The software can interpret images like a trained human expert, and sometimes, the acuity of the system is often better than a trained human expert," added Perner, who is not officially affiliated with ImageInterpret, but will be presenting research based on the software at the conference. "Our mission is that the system should give the final decision — so it's not just giving numbers, statistics, the sizes of objects, or other parameters, but the outcome is the decision a human expert would usually give — in this case, the outcome might be a certain cell pattern."
ImageInterpret started as an "application-independent" image-analysis company, meaning that its goal was to sell software to whomever needed an automated method for interpreting images in the biological sciences.
However, its flagship product was for HEp-2 cell pattern analysis, which is used in the diagnosis of autoimmune diseases. It was in this application area that the company signed its first partnership, with German firm Autoimmun Diagnostika, which markets readers and robotics for autoimmune assays.
Yet recently, like Definiens, ImageInterpret noticed the burgeoning high-content screening industry and saw the applicability of its software to fluorescence-based cell images.
Another feature that ImageInterpret is touting is its ability to analyze and interpret images of varying quality. This segues into the fact that it does not really matter what type of imaging platform is used in combination with Cell-Interpret, Perner said.
"We are not relying on image acquisition," she said. "There are a lot of different instruments around that produce images. If an assay produces an image, afterwards, the image has to get analyzed, and it doesn't matter what system produced the image. We can analyze the images from any kind of unit." She added that it is only necessary that the image is a bitmap, tiff, or some other common image file format.
— Ben Butkus ([email protected])