A procedure based on conditional probabilities to estimate proportions and densities of tissues from X-ray images of Quercus petraea samples

Abstract

This paper describes a procedure to estimate within-ring proportions and densities of earlywood vessels, fibres and parenchyma, and then validates the procedure for each tissue type. The procedure uses conditional probabilities for these estimates. It was developed using 16 annual growth rings of contrasting anatomy sampled from five mature sessile oaks (Quercus petraea Liebl.). The X-ray images of the sampled rings were converted into density maps and the two components of the rings (i.e. earlywood and latewood) were delimited manually. First, mean tissue densities of individual rings were estimated from anatomical characteristics combined with a simple analysis of the frequency histograms of density in earlywood and latewood. Second, the procedure used the radial position and density of each point of a digitised ring to calculate a probability for a point belonging to either vessels, multiseriate wood rays, fibres or axial parenchyma. The procedure was evaluated using i) measurements of tissue proportions from anatomical sections and ii) measurements of tissue densities performed on X-ray images. The validation demonstrates that the procedure provides an accurate estimation of the proportion and density of earlywood vessels, fibres and axial parenchyma in earlywood and latewood. However, it also demonstrates that the procedure is not sufficiently effective to estimate the proportion of multiseriate wood rays.