Converting our visual world into usable information and knowledge has been enhanced with new computer-aided software technologies to the great benefit of medical patients.

Definiens can also analyze images at the nanoscale as seen in this 3-D analysis of intracellular particles.

A picture, they say, is worth a thousand words. People perform this data-conversion process in every aspect of their lives to varying degrees of efficiency. But our relatively meager cognitive capabilities generally miss most of the real information contained in those pictures or visual images. Enter the world of computer-aided analysis, and the efficiency of that information-knowledge conversion process can be increased several fold, providing subtle information from those images that might normally be hidden from our view.

Researchers at Definiens, Munich, Germany, have created such a product in their recently introduced Definiens XD. This multidimensional image analysis platform has the ability to automatically compare images from different modalities, in-vivo and in-vitro data, 2-D and 3-D images over time, and 2-D images to 3-D structures in a consistent and reproducible manner. The software allows consistent and precise 3-D volumetric measurements—the most challenging but most accurate way of measuring structures of interest, including, for example, those of lesions in the human body.

The XD software is based upon Definiens’ next generation cognition network technology (CNT), an advanced technology for extracting intelligence from images. CNT was developed by Gerd Binnig, founder and head of research at Definiens and a Nobel Laureate. The underlying principle of CNT is that it is context-based. The technology differentiates itself from pixel-based and model-based matching technologies by recognizing groups of pixels as objects, picking out shapes, colors, and textures. It builds up a picture iteratively, examining objects contextually with the ability to understand scales, overlapping objects, and the relationship of 2-D images to 3-D shapes.

Definiens can be used to analyze 2-D and 3-D images in medical applications, such as the lung nodules pictured here. Photo: Merck & Co., Inc.

CNT incorporates elements from semantic networks, description logics, and functional programming. This technology can be applied to images of different magnifications, resolution, and modalities, including electron microscopy, optical microscopy, confocal microscopy, radiology, magnetic resonance imaging (MRI), conventional photography, and other sophisticated imaging variants. A unified approach with a scale- and problem-invariant processing and knowledge model is a prerequisite for the study of complex hierarchical systems, such as biological systems.

Healthcare focus

The XD software is specifically targeted at addressing the growing need for multidimensional image analysis tools in the healthcare industry, according to Wolfgang Rencken, EVP of technology and products at Definiens. The increasing need for multidimensional image analysis technologies in healthcare has been demonstrated by the growth of confocal and fluorescence microscopy imaging, kinetic and chemotactic assays, and various non-invasive imaging techniques.

Medical scientists and biologists now have the means to acquire high-content images in high-throughput scenarios. They can literally generate tens of thousands of images/day. Without some type of automated image analysis system, many of the images they create are likely to go unused due to time constraints. In medical imaging, various time-series combinations of positron emission tomography (PET), dynamic computed tomography (CT), mammography, and MRI can be used with Definiens XD to provide improved screening and treatment planning and staging.

In basic research applications, XD can be used to automatically combine fluorescent images of dyes in proteins, fixed cell fluorescence, and confocal microscopy to obtain spatial and time-series images.

And in preclinical small animal imaging, many images can be acquired at different time points from the same animal and used to study the effects of therapeutic interventions in the animals to improve the statistical comparisons of therapeutic efficacy.

While still in the initial stages of development and deployment, the XD developers have verified the accuracy of the software through comparison studies of the automated system to manual image evaluations by a panel of pathology experts. The results of these comparison studies have verified the accuracy of the XD software to be equal to or better than that of the expert analyses and considerably more consistent. More than a thousand customers have also used the system in over 200 successful scientific projects. Over the coming years, Definiens developers intend to expand its XD offerings with CAD (computer-aided detection) applications for medical professionals.

—Tim Studt