European sea bass gill pathology after exposure to cadmium and terbuthylazine: expert versus fractal analysis.

The objective of this study was to compare expert versus fractal analysis as new methods to evaluate branchial lamellar pathology in European sea bass Dicentrarchus labrax (Linnaeus, 1758) experimentally exposed to cadmium and to terbuthylazine. In particular, guided expert quantitative and fractal analysis were performed on selected images from semithin sections to test possible differences according to exposure class (unexposed, cadmium exposed, or terbuthylazine exposed) and the discrimination power of the two methods. With respect to guided expert quantitative analysis, the following elementary pathological features were assessed according to pre-determined cover classes: 'epithelial lifting', 'epithelial shrinkage', 'epithelial swelling', 'pillar cells coarctation', 'pillar cells detachment', 'channels fusion', 'chloride cells swelling' and 'chloride cells invasion'. Considering fractal analysis, DB (box dimension), DM (mass dimension), Dx (mean fractal dimension) as fractal dimensions and lacunarity from DM and Dx scan types were calculated both from the outlined and skeletonized (one pixel wide lines) images. Despite significant differences among experimental classes, only expert analysis provided good discrimination with correct classification of 91.7 % of the original cases, and of 87.5 % of the cross-validated cases, with a sensitivity of 95.45 % and 91.3 %, respectively, and a specificity of 75 % in both cases. Guided expert quantitative analysis appears to be a reliable method to objectively characterize fish gill pathology and may represent a powerful tool in environmental biomonitoring to ensure proper standardization and reproducibility. Though fractal analysis did not equal the discrimination power of the expert method, it certainly warrants further study to evaluate local variations in complexity or possible multiple scaling rules.