Chest X-ray CAD: A New Picture, A New Paradigm

	Lung CAD highlights possible nodules at University of Edinburgh, Scotland.

Lung nodules can be tricky. The ribs may obscure them, and smaller lung lesions associated with better outcomes may be too small to detect on a chest x-ray. In fact, clinical studies demonstrate that radiologists can miss 10 to 30 percent of lung nodules on chest x-rays. But CAD can change that picture.

Using lung CAD as a second reader helps radiologists detect difficult-to-see nodules. And chest x-ray lung CAD is beginning to prove itself as a helpful tool in detecting unsuspected lung nodules, potentially enabling lung cancer diagnosis in its early stages, according to recently emerging clinical studies. Chest x-ray CAD detected approximately 50 percent of nodules missed by radiologists in a study led by Joseph Jen-Sho Chen, MD, of the department of diagnostic radiology at the University of Maryland Medical Center in Baltimore.

Experts agree that there is a strong clinical case for lung CAD, particularly in the high-risk population. However, deployment is not yet a slam dunk. “Many physicians are in the wait-and-see camp,” notes Chen.

A lack of reimbursement may discourage some interested facilities from adopting the technology. In addition, providers need to iron out workflow and integration issues and understand the impact of CAD-initiated false positives before investing in the technology. As the data accumulate, early adopters continue to make the case for chest x-ray CAD.

Lung CAD and high-risk patients

Facilities across the country have started to realize the benefits of lung CAD and have integrated the technology into PACS. But since standard guidelines for the employment of lung CAD have yet to be established, the decision about when to add CAD has been left to the discretion of the physician, says Chen. Most facilities adhere to a similar model. That is, radiologists use CAD for high-risk patients, smokers or individuals being screened for cancer versus infection, explains Matthew Freedman, MD, MBA, a radiologist at the Lombardi Comprehensive Cancer Center of Georgetown University in Washington, D.C. 

CAD screening should be targeted, Freedman opines. “If you [use CAD] on patients who are at low risk, then you might find that you are picking up abnormalities, scarring or findings that are not important,” he says. “The smaller the nodule, the more likely they are to be scars in low-risk patients.”
Due to radiologist time constraints, CAD for targeted populations—especially those at high risk for lung nodule development—is a reasonable model, agrees Edwin van Beek, MD, PhD, chair of clinical radiology at the University of Edinburgh in Scotland. 

University of Maryland relies on a similar model, using chest x-ray CAD on patients considered to be high-risk, including smokers, elderly patients or those in whom the physician finds something of concern on the chest x-ray, such as a lesion, says Chen. The department’s high study volume makes it difficult to incorporate CAD into routine screening. Although it may take only several seconds more to use CAD for each chest x-ray, a study volume of a couple hundred chest studies a day could translate into a significant amount of time added to a radiologist’s work to utilize CAD, explains Chen.

CAD screening remains a work-in-progress with ongoing discussion about appropriate patient populations, sums van Beek. “For smokers, at what age do you start screening with CAD?  There are very few cancers found in those under the age 50, so that may be a reasonable area, but there is no consensus so far,” he explains.

The false-positive challenge

False-positive findings present a significant workflow issue. As CAD sensitivity increases, detection rates and false-positive rates also rise, presenting a potential workflow issue that might deter some prospective practices from investing in the technology.

“Whether you use CAD or not there are going to be false positives all the time, but I still think software can be helpful because not only does it identify possible lesions, but it can give measurements,” offers van Beek. CAD can facilitate patient follow-up by taking measurements out of the hands of the interpreter and providing nodule size changes and 3D volume. And at the University of Edinburgh, one of the most common uses for lung CAD is in the follow-up of lung nodules, including whether or not there is evidence of a reduction in size of a nodule treated with chemotherapy. 

And the numbers of false-positive findings have decreased significantly over time, notes Freedman. During the CAD development process, systems identified approximately nine false positives on each chest x-ray, compared to approximately 1.5 currently. “I expect that number to continue to improve,” he says. Despite improvements in CAD technology in terms of the declining number of false positives and earlier detection of nodules, it will still take time until physicians are comfortable with the technology, he notes. Chen is more optimistic. “It is fairly easy for physicians to dismiss the false-positive findings produced by CAD,” he says.

Countdown to deployment

The actual implementation of CAD is key, notes van Beek. “It’s a bit of dance between several people—usually a PACS vendor and a CAD company have to communicate with each other in order for the facility to be able to use CAD to the best of its ability,” he says. When CAD is deployed as an independent workstation, its use is stifled because moving between the PACS and CAD workstations  disrupts physician workflow.

In fact, workflow design is critical to successful deployment. Building CAD into the workflow after the technician obtains the images but prior to the radiologist’s interpretation, minimizes resistance and boosts acceptance, claims Chen. 

The economics of CAD

There is currently no reimbursement for chest x-ray CAD, which can make it difficult to implement the system, depending on the location, notes van Beek. “Even if it’s pennies here and there, if there is some kind of reimbursement there will be more incentive for physicians to utilize lung CAD,” adds Chen.

In the face of implementation and integration concerns, chest x-ray lung CAD could potentially reduce the cost of healthcare, as many earlier cancers are easier, and by extension, less expensive to treat than late diseases, believes Freedman.

Clinical studies of chest x-ray CAD are ongoing. The hope, says Chen, is for the technology to improve early detection of lung cancer and increase survival rates. 

Positive data could spur increased adoption and acceptance. “I think [physicians] are receptive to the idea of an unbiased system being beneficial for the patient, but better data to back it up are still needed,” sums Chen.

Lung CAD: A snapshot

The chest x-ray CAD picture remains a bit murky. Further research is needed to determine its exact benefits, and reimbursement would help sway reluctant adopters. Nevertheless, chest x-ray CAD offers a helping hand in the detection of lung nodules that may have been missed by a radiologist. In addition to boosting detection rates, lung CAD may facilitate the earlier detection of cancers, which can benefit the treatment process and potentially reduce healthcare costs. With false-positive findings declining and the technology advancing, chest x-ray CAD can be a valuable addition to high-risk screening workflow.