In contrast to emphysema, which can already be quantified automatically in CT images, manual interaction is often required for analysis of CT images with opacifications. Lung opacification occurring in diseases such as acute lung injury frequently have tissue densities (and thus CT numbers) close to that of the soft tissues of the thoracic wall, mediastinum or diaphragm sellckchem and are thus difficult to differentiate from these non-pulmonary tissues. Although initial experiences with automatic segmentation techniques of opacified lungs have been reported, none of these techniques is already available for broad experimental or even clinical use [29-31]. Consequently, the time required for manual analysis of a single whole-lung CT (median 55 slices of 5 mm thickness) can easily exceed five hours.

The potential of the extrapolation method for saving time and research resources becomes obvious when considering that the work required for manual interaction can be decreased by up to 80% when only 10 reference CT slices are analyzed. Extrapolation of adipose tissue volumes or pulmonary gas volumes has been applied by other investigators in order to limit radiation in quantitative CT studies in patients [16,17]. Our current results support this method of calculation and further underline that extrapolation is an option to simplify quantitative CT analysis. In the experimental setting, full spiral CT scans may be performed to acquire maximum information, but for the purpose of gas and tissue quantification, analysis can be limited to 10 scans.

True limitation of radiation, however, can only be achieved if 10 separate single slices of a certain thickness are “prospectively” planned and scanned one by one [17], which differs completely from spiral CT of the whole chest. Calculation of examples for effective radiation doses (for the Philips scanner) indicates that the effective radiation dose can be decreased by approximately 50% from 3.8 mSv (spiral CT) to 2 mSv (10 single slices).In our opinion, several reasons preclude the generalized recommendation to use less than 10 reference CT slices. As explained by Gattinoni et al. in a recent editorial, 10 slices with 10 mm thickness cover about 40% of the lung tissue whereas 10 slices with 5 mm thickness contain only about 20% of the lung [26].

If the number or thickness of reference slices is decreased too much, the density information Cilengitide available for extrapolation and consequently the accuracy of extrapolation decreases. As illustrated in Figure Figure2,2, analyzing 10 reference slices seems to be a reasonable compromise: bias values diverged from zero and the limits of agreement became considerably wider when less than 10 reference slices were used. As pointed out in our previous study in human patients, the adequate number of reference CT slices required for accurate extrapolation of quantitative CT results varies with the study purpose [18].