Application of automatic image analysis using a Deep Learning Neural Network for assessing the growth of green algae containing carotenoids – importance for environment, health and aquaculture

Using deep learning and neural networks enables us to greatly speed-up quantitative studies and provide a useful tool for analyzing microscopic images. Studies conducted on selected algae <i>Haematococcus</i> and <i>Coelastrum</i> sp. confirm the feasibility of using the deep learning neural network. The confusion matrix demonstrated the numbers of errors generated by the YOLO v8 network in relation to the validation dataset. It indicated a higher number of errors in the detection of <i>Haematococcus</i> than <i>Coleastrum</i>. The F1 score, as the harmonic mean of precision and recall, is significantly higher for the class <i>Coelastrum</i> sp. than for <i>Haematococcus</i> sp. Machine learning can be applied not only to the detection of individual cells, but also to the detection of colonies over a wide range of sizes. This article discussed the technical and practical aspects of implementing these advanced methods and highlighted their importance in the aquaculture, food, medical, sustainable energy, and environmental sectors.