Fully Automated CAD Application Successful in Onsite Liver Tumor Segmentation Competition

In a recent medical imaging competition, a fully automated liver segmentation application was able to correctly and consistently identify and segment 30 liver lesions representing a range of patients and pathologies.

Definiens (Munich, Germany) placed first in the 2008 Liver Tumor Segmentation Grand Challenge's live contest, organized by the International Society of Medical Image Computing and Computer-Assisted Interventions (MICCAI), held at New York University (New York, NY, USA) in September 2008.

Against a field of the world's leading research groups in the area of liver lesion segmentation, Definiens' fully automated computer-aided detection (CAD) application outperformed all other methods, including the interactive ones, and received the live competition's highest scores.

Definiens' applications accurately and consistently identified and segmented liver lesions in a series of computed tomography (CT) images from various patients. Overall, the Grand Challenge compared the brain lesion, coronary tracking, and liver-lesion segmentation techniques of 36 industry and academic teams. The performance of the liver-lesion segmentation techniques was evaluated using a set of comprehensive measures, including tumor surface area and volume.

Definiens' automatic analysis was compared against manual segmentations performed by several experienced radiologists. The success of the Definiens team in the competition followed an impressive third placing in the fully automated category of the MICCAI Segmentation of the Liver Competition 2007 (SLIVER07). Successive victories in the challenging MICCAI contests provide further validation for Definiens' pioneering semi- and fully-automated CAD applications.

"It is always a boost to receive clinical and scientific recognition for our exciting work in automated image analysis and our strong finish speaks to the innovation behind Definiens' technology,” said Frank P. Klein, VP of Medical Imaging at Definiens. "We are currently developing CAD applications for a variety of challenging cancer targets, including lung, liver, and lymph nodes.”

A striking rise in the volume of digital images to be analyzed has accompanied the increasingly prevalent use of sophisticated medical imaging systems in cancer diagnostics. Automated CAD applications can assist in overcoming the logistical logjams that are created by a shortage of clinical radiologists, and support healthcare providers in offering patients earlier, more accurate detection and treatment.

Definiens' CAD applications employ contextual information to identify lesions and tumors smaller than the human eye can detect. The company's first commercially available CAD application will be released to European customers in early 2009, and will subsequently be submitted for U.S. Food and Drug Administration (FDA) approval for the United States market.

By automating image analysis, Definiens supports healthcare providers in analyzing and interpreting vast numbers of digital images accurately and consistently. The system improves the analysis of tissue samples and non-invasive imaging, enabling translational medicine, from early diagnosis to personalized treatment.

Definiens interprets images on every scale, from microscopic cell structures to satellite images. The Definiens Cognition Network Technology is an advanced and robust context-based technology designed to fulfill the image analysis requirements of the medical, life science, and earth science markets. The technology is modeled on the powerful human cognitive perception processes to extract intelligence from images. Definiens provides organizations with fast image analysis results, allowing deeper insights enabling better business decisions.

Related Links:
Definiens