Evaluation of antifouling surfaces using a method that employs mussel larvae settlement quantified by machine learning

Antifouling coating development requires extensive performance testing. Coatings that prevent aquatic larval settlement are of interest because many forms of macrofouling begin at the larval stage. However, field testing can be time consuming and poorly controlled. Herein is reported a screening tool, Settlement of Larvae Assay using Mussels (SLAM), for down-selecting materials prior to field testing. The method entails using a dense concentration of mussel larvae that are allowed to settle on submerged test surfaces. Settled larvae are then quantified to provide a measure of antifouling performance. The SLAM test differentiated coatings with only slight differences in formulation. To enable efficient quantification of dense larvae settlement, an automated counting method was developed that combines two analyses: a color thresholding identifies larvae clumps, and a machine learning algorithm identifies non-clumped larvae. This automated 'hybrid' approach rapidly quantifies settled larvae as effectively as manual counting but in a fraction of the time.