Lung CAD Hits the Streets

What Radiologists Need To Know About Lung CAD


  • There are lung CAD options for two modalities-x-ray and CT, but it's not an either-or scenario. Either or both solutions could be justified depending on site needs. The acceptance of CT as a screening tool could play a key role in the longer term CAD market.
     
  • Both types of systems offer clinical benefits by serving as a second reader to highlight potential malignancies. CT systems provide additional workflow benefits via tools that facilitate navigation through CT data and automate clinical measurements.
     
  • Reimbursement for lung CAD is not likely in the short-term. This may not deter adoption, however, as sites that take a comprehensive look at these systems, particularly CT-based solutions, find that they are cost-effective.

Lung CAD is here, with new systems arriving to radiology departments near you. First on the scene was Deus Technologies RS2000 in 2001, a x-ray-based system designed to highlight abnormalities on chest films. Deus followed up with the RS2000D in 2003, which is also marketed by GE Healthcare as RS Digital. This second-generation system works with digital and computed radiography (DR and CR) systems or digitized films.

R2 Technology's lung CAD answer is ImageCheckerCT, a lung CAD system designed to work with CT images. And Siemens Medical Solutions offers syngo LungCARE, a viewing technology designed to aid physicians in the diagnosis of pulmonary nodules, although not a CAD system. iCAD is working on Second Look CT Lung, its CT-based CAD solution.

The rapidly evolving lung CAD market spurs a number of questions. Pablo Delgado, MD and associate professor of radiology at University of Missouri-Kansas City, explains, "CAD technology is pushing radiologists to think about things they never did before. It could initiate a dramatic change in the practice of radiology."

Lung CAD vendors tout both clinical and workflow benefits. Lung cancer is notoriously lethal; its five-year survival rate is less than 15 percent. But when lung cancer is detected in the early stages, the survival rate increases to more than 50 percent. The hitch? Only 15 percent of lung cancers are found in the early stages. Studies demonstrate that radiologists miss nodules on lung CTs in 20 to 30 percent of cases. The statistics for chest x-rays are equally compelling. Clinical studies performed for the RapidScreen system indicate that radiologists working without the x-ray CAD system miss an average of 35 percent of solitary pulmonary nodules and 42 percent of solitary pulmonary nodules between 9 and 15 mm.


X-RAY BASED LUNG CAD

Physicians rely on chest x-rays for a laundry list of indications. It's an economical procedure that is the first line of offense for most patients with lung symptoms. Deus Technologies aims to increase the yield of truly positive lung cancers by highlighting suspicious regions on chest films.

RS-2000, Deus' first generation product, works exclusively with analog films and includes a digitizer. The DICOM compatible RS-2000D (also sold by GE as the RS-Digital) works with DR, CR and digitized films. RS-Digital is a simple and efficient process. The radiologist reads the unmarked chest image. It takes 20 to 30 seconds for the CAD post-processing algorithm to analyze the digital x-ray data and merge the CAD results with the patient's electronic exam folder. The radiologist pushes a key to display the CAD results on a separate screen. Suspicious areas are circled for additional review with an average of three or fewer marks per case.

Robert Gilkeson, MD, director of cardiothoracic imaging at University Hospitals of Cleveland says RS-Digital adds an extra 5 to 10 seconds per case and has not changed workflow in the radiology department. Gilkeson is cautiously optimistic about the future of lung CAD. After using CAD for two years, he estimates that CAD has made a difference in 3 to 5 percent of cases. John Brown, MD, chairman of radiology at Garden City Hospital and a beta user of the Deus system, agrees. He notes, "There are 155,000 deaths a year from lung cancer and 170,000 new cases each year, and it's less treatable than breast cancer. If we can find small lesions and treat them early, hopefully we will have better outcomes." Indeed, early stats with Deus' systems indicate that the technology can increase detection of small lesions by up to 72 percent.

Gilkeson expects to see additional gains in true positives in the future after CAD is applied with dual-energy subtraction, which reduces obstruction of overlying bones and provides separate images for bone tissue and soft tissue. It could be one to two years before CAD and dual-energy subtraction can be applied together. In the interim, CAD sites can be optimize their investment by digitally presenting images to radiologists. Gilkeson explains, "The presentation state is important; it really helps to present information in the right way."


REDEFINING CAD WITH CT

The traditional definition of CAD is computer-aided detection. According to conventional wisdom, CAD relies on a computer algorithm to highlight suspicious areas and aid detection of malignancies. This premise behind CT-based lung CAD systems is similar to mammography CAD, but there are a few key differences. For starters, radiologists are evaluating hundreds of images with chest CTs instead of two to four x-rays, which makes workflow-enhancing tools an important part of the equation.

Randy Ernst, MD, associate professor of radiology at University of Texas Medical Branch at Galveston, was initially skeptical about lung CAD. But the changing environment with more studies per day and more images for study almost dictates assistance. Ernst explains, "We read up to 60 cases a day. Scanners are getting faster and generating more images and thinner slices. CAD never gets tired and reinforces extra care." After using CAD for more than a year, Ernst has changed his tune and says it's a very promising technology.

To further complicate the picture, lung biopsies are risky, so false positive results are more challenging than false positive mammograms. Steve Rogers, MD, chief scientific officer for iCAD, explains, "It's important to give radiologists tools to monitor change in nodules." CT-based CAD systems are designed to meet accuracy, workflow and monitoring needs. In fact, Gerald Kolb, director of economic strategy for R2 Technology, Inc. refers to CT-based CAD as 'digitally-enhanced interpretation.'

R2's ImageChecker CT Lung, iCAD's Second Look CT Lung and Siemens syngo LungCARE operate with similar objectives. The three technologies aim to improve clinical accuracy and enhance radiologist workflow and productivity. But the Siemens offering is FDA approved as an enhanced lung viewing technology that includes a second reader tool. R2 is currently awaiting final FDA signoff for the ImageChecker CT CAD system for the automatic detection of lung nodules during the review of multidetector CT chest exams.

The clinical accuracy component is simple; CAD acts as a second reader by highlighting features indicative of lung nodules. David Naidich, professor of radiology at New

York University Medical Center, explains, "LungCARE makes it much easier to evaluate abnormalities. On the detection side, our biggest concern is that we miss cancers. It is easy to miss small lesions when you are scrolling through 300 images." In fact, Brigham and Women's Hospital (Boston) reported that CAD detected clinically significant lesions in approximately 20 percent of cases where the original interpretation was normal. Other researchers report similar results. Ernst reports, "We've increased sensitivity for lung nodules by 27 percent with iCAD's Second Look CT Lung with 5 to 7mm slices. I think the results will be just as or more sensitive with 2 to 5 mm cuts."

While clinical results seem to replicate results with mammography CAD systems, (in the neighborhood of 20 percent more concerns detected) the productivity component offered by lung CAD is new. It's no secret that multidetector CT scanners yield ever-increasing amounts of data. At the same time, radiologists are in short supply. ImageChecker CT Lung includes workflow-enhancing tools that can lessen the impact of data overload and the radiologist shortage. These tools can decrease the time radiologists spend on time-consuming, mundane tasks. For example, ImageChecker CT Lung can complete automatic volumetric calculations. A radiologist can click on a region of interest to obtain automatic 3D measurements or opacity information. The system can also automatically register two data sets to measure tumor regression or progression. Preliminary studies indicate radiologists can decrease time spent reading chest CT studies by 20 percent when using the R2 workstation vs. reading axial slices. Susan Wood, PhD, vice president, CT Products for R2, concludes, "This allows the radiologist to focus time on interpretation instead of manual measurements and registration." Finally, ImageChecker CT facilitates the image distribution piece by enabling CAD results to be distributed across the enterprise as a standard DICOM file.

Second Look CT Lung features four separate modules; each module is a work-in-progress. The company expects to sell each module separately, allowing users to design a solution that best fits their needs.

QuickCue automatically detects solitary pulmonary nodules and ground glass opacities. QuickLook provides a 3D rendering of selected nodules and measures size, radiation absorption, volume and number of slices. QuickMatch automatically locates nodules in follow-up CTs and reports changes. Finally, QuickChange provides a quantitative analysis of solitary pulmonary nodules using contract-enhanced CT studies. The tool completely automates manual measurements.

Siemens aims to streamline and improve the diagnosis of pulmonary nodules with syngo LungCARE. The lung viewing technology consists of a series of tools and programs to support the visualization, evaluation and follow-up of pulmonary nodules and lesions. Naidich explains, "LungCARE addresses the spectrum of applications for evaluation of lung nodules. It allows the user to almost seamlessly scan through CT images in any plane or any slice thickness." In addition to 3D visualization of thin slices, the system also lets physicians rotate images for close-up inspection, complete automatic calculations of volume and diameter and obtain instantaneous 3D segmentations.

After the radiologist scans through the entire volume, LungCARE offers nodule enhanced viewing (NEV), a second reader tool can be used to highlight potential abnormalities. The radiologist can return to the analytic tools to re-evaluate potential abnormalities.

LungCARE stores all results in a database for future reference. When a patient returns for a second scan, the technology automatically registers both studies and finds nodules on the second study. Naidich points out, "This really saves time. Most of what we spend our time doing is comparing follow-up studies on a point by point basis."

There is a downside to CAD. These systems will locate benign small nodules, which require follow-up. Naidich continues, "As CAD systems become more and more sophisticated, they will be able to reduce the number of false positives." Currently, LungCARE detects three to five potential candidates (or nodules) per case. Most radiologists can dismiss most marks within seconds.

Naidich concludes, "The other benefit with CAD is its tremendous potential for standardization of reading." That is, there is significant variability among radiologists on which abnormalities require follow-up. CAD levels the field and applies a more uniform standard to detection and follow-up.

The upshot with lung CAD? Delgado concludes, "The evidence says that it has the potential to help us be better radiologists and take better care of patients. We are entering very promising, but uncharted, territory."





The Reimbursement Question

After (or perhaps before) users have a handle on a new technology and its clinical and workflow benefits, the next question is 'is it reimbursable.' Lung CAD is not yet reimbursed with considerations for CT-based systems differing slightly from x-ray based technologies.

H. Michael Yeh, CEO of Deus Technologies, admits, "Reimbursement is very important, but we don't have it." Yeh says Deus will continue to work to gain reimbursement, but needs more sites with usage data to formulate its case. Yeh continues, "It's not a problem to sell our system without reimbursement. It helps physicians identify more cases, and any procedures after the CAD reading are reimbursable." Hospitals can also charge patients a nominal fee to help recoup costs of the equipment. Garden City Hospital, for example, has offered x-ray CAD free of charge to patients, but plans to begin charging $20 for the procedure.

CT-based CAD exams are a different story. For starters, screening CT scans are not reimbursed. The results of the Early Lung Cancer Action Project (ELCAP) could change that by proving the utility of CT screening, but Steve Rogers, MD, chief scientific officer for iCAD, admits reimbursement will not come any time soon.

Nevertheless, CAD vendors agree that gaining lung CAD reimbursement is not as critical as securing mammography CAD reimbursement. That's because lung CAD systems incorporate workflow-enhancing tools that can increase the quality of care and decrease costs in terms of physician time.

Jerry Kolb, director of economic strategy for R2 Technology, says hospitals should consider a three-pronged approach as they evaluate the economics of CAD. For starters, if the site is not using the 3D reconstruction code, CPT 76375, it can be used and justified with lung CAD because 3D rendering is part of the diagnostic process. Next, the workflow-enhancing tools can cut a significant amount of time (and costs) from radiologists' review of chest CT scans. Finally, detected nodules do require further testing. Kolb says, "There will be increased utilization because the nodules must be followed."





State of Lung CAD: What's Available?


  • Deus Technologies - RS2000 and RS2000D / GE RS Digital-FDA approved; commercially available
  • Siemens syngo Lung Care* - FDA 510(K) cleared; commercially available
  • R2 ImageChecker CT Lung - segmentation tool pending FDA approval
  • iCAD 2nd Look CT Lung - works in progress

*system is an "enhanced lung viewing technology," not classified as CAD

Around the web

The new technology shows early potential to make a significant impact on imaging workflows and patient care. 

Richard Heller III, MD, RSNA board member and senior VP of policy at Radiology Partners, offers an overview of policies in Congress that are directly impacting imaging.
 

The two companies aim to improve patient access to high-quality MRI scans by combining their artificial intelligence capabilities.