Abstract
For morphometric analysis of desmosomes, synapses and other membrane surface features, estimating component densities on a reference surface (the membrane with which they are associated) often provides a useful practical alternative to estimating densities in a reference volume (the tissue in which they are enclosed). The present report assesses the relative efficiencies of these two approaches. In particular, the efficiencies of stereological estimators for component numerical density on a surface (Ns) and in a volume (Nv) are compared, together with those for the surface densities Ss and Sv. For this purpose, a morphometric model developed for characterizing desmosomes in human forearm skin was quantified by point and intersection counting. The efficiencies of the alternative ratio estimators were evaluated by comparing their coefficients of variation obtained on samples of equivalent size and having a similar overall analysis time. Various types of bias influencing the results are also discussed. The results demonstrate that estimators of component densities on a surface are superior to those of densities in a volume when studying epidermal desmosomes. For a given size of sample they offer a smaller coefficient of variation. In other words, they are more efficient. They may be even more advantageous in other experimental situations.