Title: Automated three-dimensional image processing for 2-blastomere and 4-blastomere embryo surgical applications
Authors: Ze Song; James K. Mills
Addresses: Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, Ontario M5S3G8, Canada ' Department of Mechanical and Industrial Engineering, University of Toronto, Toronto, Ontario M5S3G8, Canada
Abstract: The surgical tool and embryo blastomere locations are crucial to perform cell surgery. This paper presents an automated algorithm to compute blastomere centroids and surgical tool coordinate. The algorithm utilises the camera to procure Z-stack images of the workspace. It uses a three-dimensional (3D) canny filter and k-means clustering to segment and compute 3D centroid of each blastomere. For surgical tool tip location, region of interest (ROI) is established via template matching. 2D canny filter is applied on the ROI to produce segmentation results. These results are compiled together to estimate the 3D tool tip location. All computations are conducted on image plane which provide convenience to implement image-based visual servoing (IBVS) control in the future. With the overall obtained computation variations for 2-blastomere (2B) and 4-blastomere (4B) embryo, the proposed algorithm is suitable for blastomere aspiration alike procedures.
Keywords: image segmentation; computer vision; micro robotics; single cell surgery.
International Journal of Mechatronics and Automation, 2021 Vol.8 No.2, pp.92 - 99
Received: 03 Oct 2020
Accepted: 06 Oct 2020
Published online: 24 May 2021 *