Unsupervised image analysis for zebrafish embryogenesis using lab-on-a-chip embryo arrays

Abstract

The zebrafish is a popular vertebrate model organism that is widely deployed with lab-on-a-chip (LoC) technology for in-situ experiments in drug discovery and eco-toxicity assays. Although LoC devices enable easy manipulation and trapping of embryos for such experiments, constant human attention is required during and after the experiments for online and offline monitoring and analysis. The throughput and turnaround time are both limited from lack of automated image analysis. In this paper, a novel image analysis algorithm is developed to automatically recognise dead embryos and the first two stages of the zebrafish embryo development in order to detect anomalies caused by an applied chemical agent during embryogenesis. The algorithm has been examined using 55 zebrafish embryo images and has achieved a success rate of 94.5% in recognising the correct embryo development stage.