PT - JOURNAL ARTICLE AU - Rikifumi Ota AU - Takahiro Ide AU - Tatsuo Michiue TI - A novel cell segmentation method for developing embryos using machine learning AID - 10.1101/288720 DP - 2018 Jan 01 TA - bioRxiv PG - 288720 4099 - http://biorxiv.org/content/early/2018/03/27/288720.short 4100 - http://biorxiv.org/content/early/2018/03/27/288720.full AB - Cell segmentation is crucial in the study of morphogenesis in developing embryos, but it is limited in its accuracy. In this study we provide a novel method for cell segmentation using machine-learning, termed Cell Segmenter using Machine Learning (CSML). CSML performed better than state-of-the-art methods, such as RACE and watershed, in the segmentation of ectodermal cells in the Xenopus embryo. CSML required only one whole embryo image for training a Fully Convolutional Network classifier, and it took 20 seconds per each image to return a segmented image. To validate its accuracy, we compared it to other methods in assessing several indicators of cell shape. We also examined the generality by measuring its performance in segmenting independent images. Our data demonstrates the superiority of CSML, and we expect this application to significantly improve efficiency in cell shape studies.