Emphysema is a characteristic component of chronic obstructive pulmonary
disease (COPD), which has been pointed out as one of the main causes of mortality
for the next years. Animal models of emphysema are employed to study the evolution
of this disease as well as the effect of treatments. In this context, measures such as
the mean linear intercept (Lm) and the equivalent diameter (d) have been proposed to
quantify the airspace enlargement associated with emphysematous lesions in histological
sections. The parameter D2, which relates the second and the third moments
of the variable d, has recently shown to be a robust descriptor of airspace enlargement.
However, the value of D2 does not provide a direct evaluation of emphysema
severity. In our research, we suggest a Bayesian approach to map D2 onto a novel emphysema
severity index (SI) reflecting the probability for a lung patch to be emphysematous.
Additionally, an image segmentation procedure was developed to compute
the severity map of a lung section using the SI function. Severity maps corresponding
to 54 lung sections from control mice, mice induced with mild emphysema and mice
induced with severe emphysema were computed, revealing differences between the
distribution of SI in the three groups. The proposed methodology could assist in the
quantification of emphysema in animal models.
