The display of computer-aided detection (CAD) marks on individual polyps on both supine and prone CT scans improved radiologists’ detection of 6 mm-9 mm adenomas by 11 percent without impacting specificity, reported a study in the August edition of Academic Radiology.
While CT colonography (CTC) CAD has advanced in the last decade, previous studies have shown that radiologists occasionally ignore true-positive CAD marks. Current CAD systems analyze supine and prone scans independently, and the radiologist combines results with a polyp considered ‘detected’ by CAD if it is marked on either or both scans. CAD developers are not required to demonstrate detection of a polyp on both scans during software development and validation, according to the researchers.
“The purpose of this observer performance study was to determine, in a well-controlled randomized experiment, whether radiologists detected polyps more frequently when they were shown CAD marks on both the supine and prone scans rather than just on one scan,” wrote Ronald M. Summers, MD, PhD, of the imaging biomarkers and CAD laboratory at the National Institutes of Health Clinical Center in Bethesda, Md., and colleagues.
Four radiologists read 33 CT cases, 21 with one 6 mm-9 mm CAD-marked adenoma, with CAD as a first reader. Each scan had at least four false-positive marks. Radiologists read each case twice, one month apart, with one case double marked on the supine and prone scan and the other case single marked on the supine or prone scan.
Researchers used receiver operating characteristic and multiple-reader multiple-case analyses to determine sensitivity and specificity. The average per polyp sensitivities were 60 percent and 71 percent for single-mark and double-mark readings, respectively, which yielded a statistically significant increase for the average reader, the authors reported. In addition, reader confidence increased more often than it decreased for polyps found on both readings for three of the four radiologists.
The study explores a key distinction among CAD systems, Summers and colleagues said. That is, although CTC studies typically include supine and prone scans, CAD systems vary in their ability to detect polyps on both versus one scans. A system with high per-polyp sensitivity on either the supine or prone scan may not perform as well on both scans.
“Our results indicate that this distinction matters for clinical interpretation of CTC images and that there is a benefit to having CAD cue the radiologist to a potential polyp’s location on both scans,” concluded Summers, who also noted that further research is needed to determine if findings translate to concurrent or second reader paradigms.