THURSDAY, Jan. 3 (HealthDay News) -- Experienced radiologists who interpreted low-dose computed tomographic (CT) screening examinations for lung cancer had moderate to substantial interobserver agreement, though their results offered room for considerable improvement, according to research published in the January issue of Radiology.
David S. Gierada, M.D., of the Washington University School of Medicine in St. Louis, and colleagues analyzed the results after 16 radiologists reviewed image subsets from low-dose screening CT examinations of 135 participants. Interpretations could fall into one of four categories: non-calcified nodule 4 millimeters or larger, non-calcified nodule smaller than 4 millimeters, calcified benign nodule, or no nodule.
Multirater kappa values ranged from 0.58 for agreement among all four classifications to 0.64 for agreement on classification as a positive or negative screening result, according to the authors. This was similar to the agreement seen in earlier studies for mammograms and other CT interpretations. Some reader pairs showed relatively wide variation in the number of abnormalities detected and classified as positive, with differences up to more than twofold.
"Computer-aided programs that assist in the detection of lesions may improve reader performance and hold promise as a means of reducing observer variability. Semiautomated volumetric determination of lesion size may reduce variation related to nodule measurement. Further development and validation of objective, evidence-based nodule characterization criteria and automated nodule characterization algorithms also may help increase agreement at screening CT interpretation," the authors write.
Abstract
Full Text (subscription or payment may be required)
