CT scans using radiation doses as low as one-thirtieth of standard doses may be sufficient for monitoring lung nodules, a small study suggested.

When the CT scanner’s tube current-time setting was lowered from 300 milliampere-seconds (mAs) to 10 mAs, there was no loss of sensitivity for nodule detection, according to Andreas Christe, MD, of the University of Bern in Switzerland, and colleagues.

This meant that sensitivity was maintained even with a radiation dose of only 3.3% of that associated with the standard dose. This could hypothetically decrease the risk of radiation-induced cancer death from a single CT scan to one in 120,000 from one in 4,000, the researchers explained in the September issue of the American Journal of Roentgenology.

The sensitivity of CT for detecting lung nodules primarily is dependent on characteristics of the nodule, such as size and density, as well as on image acquisition parameters (tube current-time settings, attenuation).

To compare the effects of these factors on sensitivity, Christe’s group reviewed the chest CT images from 50 consecutive patients and identified nodules in 41.

The images had been obtained using tube current-time settings of 300 mAs, and reconstructed at simulated settings of 100, 40, 30, 20, 10, and 5 mAs.

Three experienced readers reviewed all images for the presence of nodules 8 mm in diameter or less.

About one-half of the nodules were expected to be solid, with a maximum attenuation of at least −300 Hounsfield units (HU), and the other half ground glass nodules with attenuation below −300 HU.

The readers found 125 nodules with a mean diameter of 3.11 mm and a mean attenuation of −300.4 HU.

A total of 39% of the nodules were smaller than 3 mm in diameter, and 53% had ground-glass appearance.

At the 5 mAs tube current level, a significant decrease in sensitivity was seen for only one of the three readers, from 82% to 77% (P=0.0035), the researchers found.

“These results suggest 10 mAs is the threshold for diagnostic image quality in this study setting and confirm the results of other low-dose CT studies,” they wrote.

Specificities at the 300 mAs dose for the three readers were 96%, 92%, and 89%, with no decreases at lower doses.

The readers differed in their ability to detect nodules according to size and density, with loss of sensitivity when attenuation was below −300 HU and size was below 3 mm, according to Christe and colleagues.

On logistic regression analysis, the primary predictor of sensitivity was density of the nodule (P

<0.0001).

The mean volume measurement error between the lowest and highest current levels was 2.2%, which was considerably less than the 38% volume measurement error between readers, they found.

The researchers also compared the nodule volumes as measured manually by the readers and as measured by computer, and found that manual reading showed larger volumes, possibly because of the inclusion of ground-glass transition areas into the diameter.

"The manual volume measurement in our study was much more influenced by inter-reader variability than by variability of tube current-time level," they observed.

Limitations of the study included a lack of data on inspiration depth and the use of simulation to construct low-dose images.

"Whether our study design represents circumstances in daily clinical routine has yet to be investigated. Results of ongoing low-dose studies without simulation should clarify the answer to this question," they concluded.

The study was supported by the Swiss National Science Foundation, the Swiss Gottfried und Julia Bangerter-Rhyner Foundation, and the Swiss Huggenberg-Bischoff Foundation.

The authors reported no conflicts of interest.

###

By Nancy Walsh, Staff Writer, MedPage Today