"Chest CT CAD is valuable to assist radiologists of all experience levels to detect lung nodules," said Dr. Charles White. "The increase in sensitivity appears to benefit most of those radiologists with the least amount of experience.
White presented the team's findings Monday at the 2006 RSNA meeting.
To evaluate the use of CAD as a second reader to detect lung nodules on chest CT scans, the researchers obtained 110 cases from four different sites. The study population included 64 women and 46 women with an average age of 58.6. There were 23 smokers, 63 nonsmokers, and 24 with unknown smoking status.
Nine cases were produced from a four-slice CT scanner (GE Healthcare, Chalfont St. Giles, U.K.), 69 cases were contributed by a 16-slice CT scanner (Philips Medical Systems, Andover, MA), and 32 cases were generated by a 40-slice system (Philips). Slice thicknesses varied from 0.9 mm to 3 mm, and there was variation in both acquisition protocols and clinical indications for the exams, White said.
Inclusion criteria included a technically acceptable quality of a clinically indicated chest CT scan, 10 or fewer actionable nodules, slice thickness less than 3 mm, and a contiguous or overlapping scan, according to White.
In the first part of the study, three radiologists with more than 10 years of experience in reading chest scans read the studies, and identified all nodules between 4 mm and 30 mm in maximum diameter. The 108 total nodules identified by all radiologists were used as the reference standard for the study.
CAD software (Philips) was then applied. Three radiologists with varying levels of experience then reviewed studies both with and without CAD findings.
The first reader (less than two years of experience reading chest CT scans) found 69 nodules (63.9%) without CAD and 77 (71.3%) with CAD, while the second reader (greater than two years but less than five years of experience) found 73 nodules (67.6%) without CAD and 83 (76.9%) with CAD. The third radiologist (more than five years of experience reading chest CT scans) identified 82 (75.9%) without CAD and 86 (79.6%) with CAD, according to White.
The difference in sensitivity was only statistically significant (p < 0.05) for the first and second readers, White said.
White acknowledged limitations of the study, including its reliance on one reader for each level of experience.
By Erik L. Ridley
AuntMinnie.com staff writer
November 27, 2006
Chest x-ray lung CAD reimbursement picks up, October 24, 2006
Barriers still hinder lung CAD adoption, August 24, 2006
Pulmonary nodule decisions must be individualized, August 16, 2006
Canadian lung CAD cuts false positives, July 24, 2006
CAD improves lung nodule detection, June 1, 2006
Copyright © 2006 AuntMinnie.com