Abstract
The classification of coralline algae commonly relies on the morphology of cells and reproductive structures, along with thallus organization, observed through Scanning Electron Microscopy (SEM). Nevertheless, species identification based on morphology often leads to uncertainty, due to their general plasticity. Evolutionary and environmental studies featured coralline algae for their ecological significance in both recent and past Oceans and need to rely on robust taxonomy. Research efforts towards new putative diagnostic tools have recently been focused on cell wall ultrastructure. In this work, we explored a new classification tool for coralline algae, using fine-tuning pretrained Convolutional Neural Networks (CNNs) on SEM images paired to morphological categories, including cell wall ultrastructure. We considered four common Mediterranean species, classified at genus and at the species level (Lithothamnion corallioides, Mesophyllum philippii, Lithophyllum racemus, Lithophyllum pseudoracemus). Our model produced promising results in terms of image classification accuracy given the constraint of a limited dataset and was tested for the identification of two ambiguous samples referred to as L. cf. racemus. Overall, explanatory image analyses suggest a high diagnostic value of calcification patterns, which significantly contributed to class predictions. Thus, CNNs proved to be a valid support to the morphological approach to taxonomy in coralline algae.
Highlights
They are common in Mediterranean benthic communities, constituting biodiversity hotspots known as maerl beds and coralligenous habitats [2,5,6]
We presented a new classification tool for coralline algae diagnosis, by applying a deep learning technique to Scanning Electron Microscopy (SEM) images for the automated identification of four species at different taxonomic levels; We developed and evaluated Convolutional Neural Networks (CNNs)-based classification models against two baselines, namely a dummy classifier and a human-reported classification
Our model was tested in a practical scenario, to support the classification of two uncertain samples of coralline algae; We investigated and discussed the contribution of six main morphological categories, shown in the SEM images, to the classification task; We explored a set of explanation methods, which justify the class assignment of the proposed model by visually highlighting the contribution of portions of the processed were enriched by processing each image together with the vectorized representation of the observed morphological features
Summary
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. Modarres et al [38] used CNN for SEM image recognition of generic nanostructures, mostly of non-biological origin Given their morphological plasticity, new diagnostic tools for species identification could provide significant support to the experts, especially for fossil samples with poorly preserved morphological features. We summarize the contribution of this research as follows: We presented a new classification tool for coralline algae diagnosis, by applying a deep learning technique to SEM images for the automated identification of four species at different taxonomic levels; We developed and evaluated CNN-based classification models (open-sourced on GitHub as reported in the Data Availability Statement) against two baselines, namely a dummy classifier and a human-reported classification. Discriminate L. pseudoracemus from all the other species (2 class−CNN); Classify the three genera (3 class−CNN); Classify the four species (4 class−CNN)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
