Quantitative assessment provides valuable tool to crunch data

June 16, 2005

Automated numerical analyses of medical images can play a key role in radiological research and diagnosis. These techniques are emerging as a powerful means of improving diagnostic efficacy and efficiency, according to speakers at the European Congress of Radiology in March.

Automated numerical analyses of medical images can play a key role in radiological research and diagnosis. These techniques are emerging as a powerful means of improving diagnostic efficacy and efficiency, according to speakers at the European Congress of Radiology in March.

Quantitative assessment is likely to become an integral part of clinical trials of novel interventional radiology approaches, said Prof. Johan Reiber, a professor of medical imaging at Leiden University Medical Center in the Netherlands. Trials investigating the effect of implanted devices often involve detection of tiny changes from baseline measurements. Even experienced medical specialists can struggle to make this determination with a high degree of diagnostic confidence.

"Manual caliper analysis techniques are time-consuming and remain very observer-dependent. As a result, clinical trials become unnecessarily large and expensive, and outcomes are delayed," he said. "We need better techniques, and computers can help."

The technique known as quantitative coronary arteriography (QCA), already widespread in interventional cardiology trials, is now being developed for radiologic applications. QCA is generally used to map blood vessel diameter between two fixed points, which yields important information on vessel obstructions prior to and following stent placement. Researchers are fashioning a more sophisticated approach for interventional radiology because treated coronary or carotid vessels may contain side branches or bifurcations (Figure 1).

"We cannot just do a separate straight analysis on the different branches of the vessel. We need to have a complete understanding of the vessel architecture," Reiber said.

Computerized measurement tools can also aid cardiovascular MR (Figure 2). Comprehensive imaging studies produce hundreds or even thousands of images that radiologists must study. An automated analysis tool capable of mapping the heart could help create a computer-aided diagnosis package for cardiac function.

The usefulness of quantitative assessment is not restricted to morphology alone. Researchers at Leiden are designing software for vessel wall imaging, using 64-slice CT angiography, that identifies and characterizes lesions in the coronary tree in addition to mapping lumen morphology. Further investigation, including clinical trials, should reveal the value of quantitative CTA in risk stratification of postinfarction patients.

"Clinicians need the total coronary plaque burden; that is, the number and localization of lesions and also the composition of flow and nonflow data," Reiber said. "Quantitative CTA will be very important for screening purposes and for risk assessments. We need to find which treatment is suitable for which patient."

Quantitative imaging should be regarded as the next standard for radiology, following structural imaging, functional imaging, and molecular imaging, said Prof. Joseph Ferrucci, emeritus professor and radiology chair at Boston University School of Medicine. Ferrucci heralded the rise of numerical interpretation in his inaugural lecture at ECR 2005.

"Quantitative imaging techniques represent a new synergy between radiology, physics, and computer science and may represent a new subdiscipline in radiology," he said.

CAD is a prime example of a quantitative imaging technique that is coming into its own, according to Ferrucci. A number of CAD algorithms have been developed to assist in the early detection of lung and breast cancer. They work by comparing numerical data about lesion morphology with statistics and biofeedback. Other targets, such as pulmonary emboli, signs of multiple sclerosis, and atherosclerotic plaque, are under evaluation, although cancer screening is likely to remain the prime application for CAD.

"The problem with cancer screening by imaging is that we have huge data sets to analyze in which we are looking for rare, random, and potentially lethal findings," he said. "The solution is a computer-based data preprocessing tool that gives us improved sensitivity, reader speed, and confidence."

Cancer screening programs may also benefit from another quantitative imaging technique known as volumetry. Doubling time is a critical parameter in assessment of lung cancer lesions. Manual measurements of tumor diameter from CT slices are not only time-consuming but error-prone. Automated calculation of tumor volume from the whole CT data set offers a faster and more reliable means of assessing lesion growth.

"A small increase in diameter growth can correspond to a much larger volume increase, and it is for this reason that volumetry has become so exciting as a new radiological technique," Ferrucci said. "With a single mouseclick on a liver metastasis, you can make a measurement, show the volume, and get a printout, and this can all be done reproducibly and rapidly."