Vendors target cancer with software designed to visualize lung nodules

April 2, 2003

CAD package and new visualization analysis tools debut in EuropeTwo additions to the lung CT software market debuted at the European Congress of Radiology, held March 7 to 11 in Vienna. Philips Medical Systems introduced its new CT

CAD package and new visualization analysis tools debut in Europe

Two additions to the lung CT software market debuted at the European Congress of Radiology, held March 7 to 11 in Vienna. Philips Medical Systems introduced its new CT Lung Nodule Assessment (LNA) and Comparison option, a software program that can be added to all scanners in the company's Mx8000 line. Meanwhile, R2 Technology used the European meeting to launch its ImageChecker CT LN-1000 system, a combination of hardware and software tools with computer-aided detection functionality.

Philips received FDA clearance for its LNA program just weeks before the meeting, while R2 used the show to announce the first worldwide beta installation of its lung CT CAD software at the nearby University of Vienna. FDA approval of CAD functionality on the ImageChecker CT workstation is still pending, though the enhanced visualization and review tools are cleared for sale in the U.S.

R2's complete ImageChecker CT LN-1000 includes a workstation set up to receive thoracic multislice CT exams and DICOM-structured reports of areas identified by CAD. On reviewing the CT data, radiologists can access a number of options to enhance suspected pulmonary nodules. These include LungMap, for navigation and nodule visualization, and LungView3D, which displays nodules and associated anatomy with automatic vessel and wall segmentation.

Likely cancers are colored in green. Findings related to suspicious nodules, including automated measurements of volume and area, can then be saved for comparison in follow-up examinations. A temporal comparison option for automated tracking of individual nodules is being developed as a work-in-progress.

"We have 3D data out of CT, so we can actually see the lung nodule in multiple dimensions. We can separate it from the background, so it gives us a very accurate way of detecting small lesions, down to about 4 mm," said Edward Barker, R2's director of strategic sales. "Because it's a computer, it doesn't care whether it looks for one nodule or 50 nodules. It'll mark them all, you can record them all. When the patient returns in three months, all you have to do is look at each of those nodules again and see whether they have changed."

Philips' LNA program offers a similar range of visualization and review tools, but with temporal comparison already in place and without the CAD functionality. Users can segment identified lung nodules with a single mouse click, report quantitative data, and then mark the nodule for later reference. Clinical results generated from the LNA package can be automatically reported-electronically as a PDF file or on paper-directly from the MxView Visualization System. When the patient returns for a follow-up exam, previous images are reloaded automatically and nodules matched for comparison.

Software packages that enhance the visualization of lung pathologies promise to assist the prompt, accurate diagnosis of lung cancer. These high-tech analysis tools help identify suspect lung lesions early and track their progression over time.

R2's interest in thoracic imaging marks a major step in the company's aim to broaden its focus from mammography to additional anatomical areas and modalities. The ImageChecker CT lung software showcased at the ECR is built on R2's OmniCAD technology, a universal platform with open architecture. Previewed at the 2002 RSNA meeting, the OmniCAD platform is intended to provide a vendor-neutral means of supporting image review and CAD across a range of modalities and clinical applications, including cancers of the lung, breast, and colon and pulmonary embolism.

"Once you've made some algorithms and built a platform that can detect cancer, you just need to teach it what cancer looks like in those other organs," Barker said.

For Philips, commercial release of its lung CT software highlights successful integration and development of its CT business, acquired through the purchase of Marconi Medical Systems, according to Thomas van Elzakker, director of global marketing strategy for CT radiology products.

"We are already showing applications that are the product of the CT group working together with the Philips' research center in Hamburg," he said.

Both companies are also keen to point out the clinical advantages to be reaped from early detection and more focused follow-up of pulmonary nodules, made possible with lung CT software. Lung cancer has the highest incidence and mortality of all cancers, accounting for an estimated 900,000 deaths worldwide every year. Incidence is believed to have increased by 16% over the past 15 years, with average five-year survival rates put at 14% for the U.S. and 8% for Europe.

With the advent of multidetector technology, CT has become a viable option for lung exams, offering fast, high-quality imaging to patients with respiratory difficulties. Timely processing of the vast quantities of data generated, however, is a key issue.

"You have the ability now to get a lot of data and at higher resolution, because you have thinner slices. But the same radiologist now has to read twice or three times as much data. And guess what-you don't get fewer patients, you get more," Barker said.

R2 began taking orders for the ImageChecker CT LN-1000 at the ECR. Production is expected to begin in April, once beta testing is complete. In the meantime, radiologists at hospitals in France and Italy who participated in development of the software are being encouraged to join their Austrian colleagues in testing the prototype. Newly signed alliances between R2 and Vital Images (SCAN 2/19/03) and Sectra Imtec, the latter deal also announced during the ECR, will likely aid product dissemination once the software is ready to ship.

R2 must still convince the FDA to approve the CAD functionality on its lung CT software. While European radiologists will soon be able to purchase the complete ImageChecker CT LN-1000 for lung nodule detection, the system remains limited to investigational use in the U.S. In practice, this means workstations shipped to U.S. radiologists will arrive equipped with all advanced visualization, review, and measurement tools. Users will not, however, be shown possibly suspect lesions that have been identified by a CAD algorithm that circles them in red.

Philips, however, will be marketing the exact same version of its CAD-free software on both sides of the Atlantic. Although the company has been working on a smart system to flag suspicious nodules as well, there are no plans to add this feature to the software package just yet.

"It is very important to prove that the CAD portion helps and is not counterproductive to healthcare in general. The FDA makes that clear as well," van Elzakker said. "We don't want to bring in another commercial option until we're pretty sure that it does what it needs to be doing."

This cautious approach to automated detection for lung CT is shared by GE Medical Systems, whose CT Advanced Lung Analysis tool similarly offers automated measurement, nodule matching, and 3D image reconstruction, but again without the CAD label. Similarly, the PrimeLung package from AccuImage, which gained FDA 510(k) clearance just a few weeks ago (SCAN 3/5/2003) offers modules for image review, manipulation, and data generation only. (The SCAN headline erroneously referred to lung CAD.) An automated nodule detection module will be submitted separately to the FDA once it gets through clinical trials. CADx Medical Systems also plans to launch CT products for the lung and colon, but not until 2004.

Little wonder, then, that vendors not offering CAD functionality are emphasizing the technological superiority of their software visualization, measurement, and workflow management tools. For Philips, for example, it's a question of accuracy and smart features to assist follow-up.

"There is not much difference in the speed offered by the different providers," van Elzakker said. "But we have the ultrahigh resolution reconstruction that others don't have, and I think what really makes a difference is the combination of functional IT elements in the package."