Retrospective review of CT lung screening signals need for advanced image processing

Lung cancers discovered on annual repeat screenings were often identified in the previous round of screening, which suggests the incorporation of advanced image-processing techniques and additional display methods could boost early detection, according to a study published online in American Journal of Roentgenology.

Authors Dong Ming Xu, MD, and colleagues from the Department of Radiology at Mount Sinai School of Medicine in New York City also suggested a review of the varied appearance of malignant nodules could also be beneficial.

“Understanding the reasons for not identifying a malignant nodule in the context of screening will help in finding lung cancer earlier,” they wrote.

Xu and colleagues reviewed CT images of patients with lung cancer diagnosed in annual rounds of screening as part of the International Early Lung Cancer Action Program (I-ELCAP). Three radiologists reviewed the scans of 104 lung cancer patients, assigning findings to one of three categories: 1, cancer was not visible at previous CT screening; 2, cancer was visible at previous CT screening but not identified; 3, abnormality was identified at previous CT screening but not classified as malignant.

Results showed that 80 (77 percent) of the patients in the study could have had their cancers identified on the previous screening. These patients all had clinical stage I disease on a prior screening image, though by the time the subsequent annual screening occurred, the disease had progressed beyond stage I in 14 patients (18 percent).

Xu and colleagues noted that among the 56 patients in category 2, 17 had nodules that were smaller than 3 mm, which would not have been referred for further workup even if they had been identified. Two of these patients, both with small cell carcinoma, had disease progression beyond stage I by the following year.

The authors pointed out that 21 patients had nodules larger than 3 mm that were similar in size to adjacent blood vessels, and one-third of these patients had their disease progress beyond stage I by their next screening. “This finding suggests that use of computer-assisted diagnosis, which is particularly useful for separating nodules from blood vessels, would have led to even earlier diagnosis,” wrote Xu and colleagues. “Perhaps in the future such visualization techniques will become an integral part of the reading process.”

Until these techniques are implemented, the authors suggested additional imaging in maximum intensity projection may facilitate discrimination between blood vessels and nodules due to the fact that these images show a longer portion of the vessel than conventional images.

Twenty-four patients had category 3 findings in which an abnormality was found on previous screening but not classified as malignant. Xu and colleagues speculated that this may be due to lack of experience in identifying small cancers with an unusual appearance. These nodules were misclassified as fibrosis, scarring and simple bulla until growth was seen. They noted distinguishing features—convex borders, marked asymmetry of apical scarring, the presence of small cystic airspace with irregularly thickened walls—which should alert the radiologists that follow up is needed.

“We believe these cases have educational value for the imaging community and are working to provide a subset of them in a public database,” wrote the authors. “We hope that other large screening programs will contribute to such databases.”

Evan Godt
Evan Godt, Writer

Evan joined TriMed in 2011, writing primarily for Health Imaging. Prior to diving into medical journalism, Evan worked for the Nine Network of Public Media in St. Louis. He also has worked in public relations and education. Evan studied journalism at the University of Missouri, with an emphasis on broadcast media.

Trimed Popup
Trimed Popup