Fractal analysis of alveolarization in hyperoxia-induced rat models of bronchopulmonary dysplasia.

Abstract

No papers are available about potentiality of fractal analysis in quantitative assessment of alveolarization in bronchopulmonary dysplasia (BPD). Thus, we here performed a comparative analysis between fractal [fractal dimension (D) and lacunarity] and stereological [mean linear intercept (Lm), total volume of alveolar air spaces, total number of alveoli, mean alveolar volume, total volume and surface area of alveolar septa, and mean alveolar septal thickness] parameters in experimental hyperoxia-induced models of BPD. At birth, rats were distributed between the following groups: 1) rats raised in ambient air for 2 wk; 2) rats exposed to 60% oxygen for 2 wk; 3) rats raised in normoxia for 6 wk; and 4) rats exposed to 60% hyperoxia for 2 wk and to room air for further 4 wk. Normoxic 6-wk rats showed increased D and decreased lacunarity with respect to normoxic 2-wk rats, together with changes in all stereological parameters except for mean alveolar volume. Hyperoxia-exposed 2-wk rats showed significant changes only in total number of alveoli, mean alveolar volume, and lacunarity with respect to equal-in-age normoxic rats. In the comparison between 6-wk rats, the hyperoxia-exposed group showed decreased D and increased lacunarity, together with changes in all stereological parameters except for septal thickness. Analysis of receiver operating characteristic curves showed a comparable discriminatory power of D, lacunarity, and total number of alveoli; Lm and mean alveolar volume were less discriminative. D and lacunarity did not show significant changes when different segmentation thresholds were applied, suggesting that the fractal approach may be fit to automatic image analysis.