Ultralow-dose lung MDCT may still find ground-glass lesions By Eric Barnes AuntMinnie.com staff writer November 30, 2005
CHICAGO - Thoracic radiologists pay special attention to the so-called ground-glass opacities (GGOs) they find in some patients' CT lung images, and for good reason. The generally nonspecific findings, seen as mottled areas of elevated attenuation, may be associated with various interstitial and bronchoalveolar pathologies, and thus a potentially higher risk of malignancy.
In practice, the fear of missing GGOs is one of the reasons why some radiologists are reluctant to embrace ultralow-dose CT lung screening.
At yesterday's RSNA lung screening sessions, researchers from Japan addressed the issue by presenting findings from their study using a commercially available GGO phantom, which they subjected to various CT doses to determine the minimum exposure levels needed to detect GGOs in CT datasets. The results for low-dose screening advocates are encouraging.
Yoshinori Funama, Ph.D., from the University of Kumamoto in Japan presented his group's research, aimed at determining the low-dose limits for GGO detection at CT.
"Ground glass accounted for 91% of missed lesions at low-dose screening CT" in one study, Funama said.
The group repeatedly scanned a commercially available chest phantom with acrylic simulated ground-glass opacities on a 40-detector-row scanner (Brilliance 40, Philips Medical Systems, Andover, MA).
Collimation was set at 1.25 mm, rotation time was 0.5 sec, and the reconstructed slice thickness was 5 mm at 120 kVp. The kVp remained constant while the mAs settings were varied during each scanning sequence at 15, 22.5, 90, and 180 mAs. Attenuation values of the simulated lung parenchyma and GGOs were -900 and -800 HU, respectively.
"Simulated GGOs sized 4, 6, 8, 10, and 12 mm were located at levels of the lung apex, the bifurcation of the trachea, and the lung base," Funama said. "We extracted 10 x 30-mm rectangular areas with or without GGOs from the chest phantom images scanned with each mAs (setting). Then we prepared a total of 450 images (five different mAs settings x 30 images with and without GGOs x three scan levels) for the observer performance test."
Diagnostic performance was estimated using Az values, expressed as the area under the ROC curve, and CTDIw dose index was calculated using a cylindrical phantom, he said.
The results showed Az values of 0.873 at 15 mAs, 0.892 at 22.5 mAs, 0.905 at 45 mAs, 0.894 at 90 mAs, and 0.908 at 180 mAs. The overall difference in the Az values of the five datasets was not statistically significant (p = 0.43), according to a two-way ANOVA test. The measured CTDIw was 1.2 mGy at 15 mAs, 1.8 mGy at 22.5 mAs, 3.7 mGy at 45 mAs, 7.4 mGy at 90 mAs, and 14.7 mGy at 180 mAs.
"Detectability of GGOs at 45 mAs or 90 mAs compared (favorably) to that achieved with 180 mAs," Funama said, while "the CTDIw at 45 mAs was reduced by 75% relative to that at 180 mAs. Detectability of GGOs was preserved at scan with 15 mAs as well as at scan with 180 mAs."
Nevertheless, noise was higher in the upper portion of the lung phantom compared to the middle and lower regions, he said, which presents an important issue for future research. Noise reduction filters may be helpful, Funama said.
A separate report presented yesterday by the Radiology-Pathology Consensus Group (RSNA abstract SSG04-01) examined 125 cases of previously diagnosed broncholoalveolar cell carcinoma (BAC) to correlate CT attenuation with histopathology. Among 12 pure ground-glass nodules in the cohort, showing varying degrees of infiltration, only one was unanimously agreed upon as pure BAC by the pathology panel, the group wrote. The researchers concluded that pure BAC was rare, and found a strong correlation between nodule attenuation and the degree of invasion.
By Eric Barnes
AuntMinnie.com staff writer
November 30, 2005