Human lung volume, alveolar surface area, and capillary length.

We compare the effectiveness of morphometric methods for estimating lung parameters. Various stereological methods are applied on human lungs and described in detail. The lung volume was estimated by Cavalieri's principle and by fluid displacement. Both methods are reliable, but Cavalieri's principle is superior when systematic sections are needed or when volumes of parts of the lung are wanted. Point counting demonstrated that 87.5% of the lung is parenchyma, 5.4% is vessel volume, and 7.1% is bronchia volume. Alveolar surface was estimated on vertical and isotropic uniform random tissue (IUR) sections. The capillary length and length density was estimated on projected images of vertical slices (Gokhale method) and on IUR sections. Only minute differences were found whether IUR sections or vertical sections were used. Of the total variation, approximately 2% was due to the stereological variation and approximately 98% was due to the biological variation on IUR sections and vertical sections. Estimates for volumes, surfaces, and lengths coming from model-based and design-based methods gave similar results for human lungs. In our hands, the design-based methods were easier to use and required less time. However, only the design-based methods offer the guarantee of an unbiased estimate.