Detection and Classification of Zebrafish Embryonic Phenotypes to Signaling Pathways using Enhanced YOLOv8

The ability to accurately detect and classify zebrafish embryonic phenotypes is crucial for understanding developmental biology and evaluating the effects of genetic and chemical perturbations on signaling pathways. This study presents an advanced deep learning-based framework using YOLOv8 for automated identification and categorization of the anatomical features and developmental phases of zebrafish. A dataset is used to train the model. Comprising 17 distinct phenotype categories, including Abdomen, Blastomere, Dead, Epiboly, Eye, Head, Somite, Tail, Yolk, and multiple developmental stages (Stage1–Stage7, Tail Extension, etc.). The YOLOv8 architecture is fine-tuned for high-precision phenotypic identification, leveraging its real-time object detection capabilities and improved feature extraction mechanisms. The system enables rapid and accurate segmentation of embryonic structures, facilitating quantitative assessments in biomedical research, genetic screening, and drug discovery. The model's resilience and effectiveness are confirmed by classifying zebrafish phenotypes with high accuracy.