Towards automatization in morphometric analysis of sea urchin embryos for ecotoxicological applications and near-future scenarios simulation

Ecotoxicological bioassays are widely recognized as excellent tools for detecting the bioavailability and toxicity of many environmental pollutants. Specifically, embryo bioassays with invertebrates are among the most sensitive approaches used in the ecological quality assessment of the marine environment. However, these tests are time-consuming and expert judgment dependent, which potentially affects the results based on the ability of the operator and the number of counted embryos. These limitations can be overcome by developing a morphometric analysis based on image acquisition. Herein, a completely automated acquisition system of images for a fast and effective discriminant morphometric analysis is faced by using digital pictures of embryos of the Mediterranean Sea urchin Paracentrotus lividus. Linear Discriminant Analysis (LDA) was applied to find the best combination of automatically acquired morphometric parameters capable of discriminating among different morphological phenotypes by following the expert judgment. The automatized method shows a performance of 82% in detecting normal embryos vs a performance of manual observation of 79%. Our findings highlight the parameter area/perimeter ratio as the most critical descriptor to discriminate among different morphological phenotypes based on a predetermined classification covering six morphotypes. The method was validated on embryos exposed to elutriates from contaminated marine sediments. The high proficiency in classifying the embryos reveals the suitability of this method for the future assessment of marine pollution and complex scenario simulations.