Automated 3D blastomere segmentation and injection for early-stage embryo

Abstract

Automation of single biological cell surgery requires that the location of organelles and cell structures to be determined to permit automated processing to be carried out. In this work, z-stack images of mouse embryos are used to develop image processing algorithms to determine the position (x, y, z) coordinates of the embryo blastomeres. These images are processed utilising existing Matlab image processing algorithms. The embryos are transparent, allowing a series of images along the vertical cell axis to be obtained. Individual z-stack images are processed using 2D image processing steps to segment, then estimate the centroid (x, y) coordinates of the blastomeres in the 2D image. Successive processing of all z-stack images permits the centroid of the blastomeres to be determined in (x, y, z) coordinates. Image processing-based calibration allows a PZD micropipette to move to the computed centroid positions under closed-loop control. These algorithms are verified using z-stack images of mouse embryos at the two-blastomere stage.