Image analysis helps speed device and drug product trials

November 6, 2002

Technology may eventually transition to therapyThe evolution of expert software systems has opened new areas for the application of medical image analysis. Computer-aided detection systems have already found a place in breast tumor

Technology may eventually transition to therapy

The evolution of expert software systems has opened new areas for the application of medical image analysis. Computer-aided detection systems have already found a place in breast tumor screening and appear to be moving into lung screening as well. But not all such applications of technology pertain to radiological practice. A company spun off from academia has developed a profitable business around proprietary MR and CT volumetric analysis technology as the means to test drugs and devices for use in osteoarthritis, oncology, and neurological applications.

VirtualScopics of Pittsford, NY, was formed in 1999 to develop and commercialize automated image analysis software first developed in a cooperative effort between medical and engineering faculty at the University of Rochester. The most significant aspect of VirtualScopics' image analysis technology, according to Mikael Totterman, the firm's chief operating officer, is its ability to automatically segment tissue structures, particularly soft-tissue features such as the cartilage in knee joints. This software-driven segmentation can be applied to low- as well as high-contrast images.

"Some of these items have been very difficult or near impossible to segment consistently in the past," he said.

The Rochester scientists and engineers developed their analysis software as part of a multiyear effort to "encode the radiologist's brain," Totterman said. All the various factors involved in the identification of a particular structure, such as contrast, texture, shape, and location, were translated into computer algorithms in order to repeat the process accurately and reproducibly.

The software is able to measure the volume of a tumor in an MRI data set to within 2% accuracy. This analysis can then be reproduced on future scans to monitor the progress of tumor shrinkage in response to treatment. Similarly, the VirtualScopics technology is being used to track the impact of therapy on arthritic joints and neurological conditions, including multiple sclerosis and Alzheimer's disease.

VirtualScopics initiated a strategic partnership with pharmaceutical giant Pfizer last summer to provide analyses of the progress of drugs undergoing clinical testing for oncology, osteoarthritis, and other uses. The objective is to allow the testing firm to decide more quickly whether to go forward in the development of particular therapeutic compounds, thus focusing development resources more effectively. Pfizer, which has made an equity investment in the small venture capital-backed private company, is also helping to validate the efficacy of the analysis technology, Totterman said.

Along with lining up additional pharmaceutical partners, the company is working with manufacturers of surgical implants, such as hip prostheses and other devices, again to speed the decision-making process in clinical testing, he said. Eventually, VirtualScopics' work in monitoring drug therapies and implants may lead to clinical applications of the technology, although this is a long-term process requiring FDA certification.

"If these drugs do get approved, there will probably be a similar need to track them and see if individual patients are responding to the therapy as expected," Totterman said. "Being able to provide this type of information on an individual patient basis may have a lot of value in helping to manage the therapy itself."