RSNA: Colon CAD makes its mark
CHICAGO--Colon computer-aided detection (CAD) "can deliver" was the message of Abraham Dachman, MD, of the University of Chicago Hospital, on Monday at the annual Radiological Society of North America (RSNA) meeting. Dachman shared the results of a meticulously designed research study intended to assess the impact of CT colonography CAD on physician sensitivity and specificity for detection of polyps in a multi-reader, multi-case trial.
To date, numerous stand-alone studies have demonstrated the sensitivity of CAD, said Dachman. Widespread acceptance, however, hinges on reader studies. The large, multi-reader, multi-case study focused on iCAD's CT Colon CAD software, which had already cleared the sensitivity bar of stand-alone CAD studies.
The trial consisted of 100 screening cases and involved 19 board-certified radiologists from both academic and community settings. Each reader performed 200 readings, reading each case twice.
The results are quite promising. The sensitivity of the average reader improved by segment, by patient and by polyp. In fact, reader sensitivity improved by about 5.5 percent for patients with both small and large polyps with the use of CAD, with small adenomas seeing the greatest gain in sensitivity.
Dachman did acknowledge a slight decrease in specificity for some readers with the use of CAD; however, overall specificity remained over the 90 percent mark. In addition, reading time increased slightly, and the number of false positives also nudged up slightly.
To date, numerous stand-alone studies have demonstrated the sensitivity of CAD, said Dachman. Widespread acceptance, however, hinges on reader studies. The large, multi-reader, multi-case study focused on iCAD's CT Colon CAD software, which had already cleared the sensitivity bar of stand-alone CAD studies.
The trial consisted of 100 screening cases and involved 19 board-certified radiologists from both academic and community settings. Each reader performed 200 readings, reading each case twice.
The results are quite promising. The sensitivity of the average reader improved by segment, by patient and by polyp. In fact, reader sensitivity improved by about 5.5 percent for patients with both small and large polyps with the use of CAD, with small adenomas seeing the greatest gain in sensitivity.
Dachman did acknowledge a slight decrease in specificity for some readers with the use of CAD; however, overall specificity remained over the 90 percent mark. In addition, reading time increased slightly, and the number of false positives also nudged up slightly.