IEEE Journal of Selected Topics in Quantum Electronics(2023)
引用0|浏览5
暂无评分
摘要
Caenorhabditis elegans
is widely used as a model organism and its oocyte and embryos’ reproductive development are under extensive study due to their ease of observation and cultivation. As one of the main embryo constituents, yolk granules and their motion serve important functions such as storage of energy and materials. However, current observation and analysis methods have issues such as being toxic, problematic identification of densely distributed particles, or being too slow to track fast movement. Here we use a home-built, label-free imaging system that faithfully images the particles at a high frame rate. Then, a trained U-Net with an upsampling block is used to recognize the granules with high accuracy compared to traditional methods (83% vs 50%). Through motion analysis, we found that most granules can only be tracked in a short period of time (thus presenting an additional challenge for slower imaging methods). While typical diffusion model is not appropriate for short tracks, we use the track-averaged speed and its distribution parameters to characterize the intra embryonic motions, classifying embryos from normal and starved mothers, thus demonstrating that our method can be used to quantitatively evaluate the embryo's quality without any extraneous labels.