Machine vision guided robotics for automated microinjections into fruit fly embryos

PROJECT SUMMARY Objective Biotechnology is a spin-out from the University of Minnesota that is pioneering the use of machine vision guided robots for automating precise microbiology procedures such as microinjection. Drosophila melanogaster (fruit fly) is a model organism extensively used in both basic and clinical research. A key challenge

in Drosophila biological research is the bottleneck of generating and maintaining transgenic lines of flies. Traditionally, transgenesis involves skilled technicians performing intricate and precise microinjection procedures repeatedly. However, Objective Biotechnology Inc. has introduced an innovative solution - a

machine vision guided robot designed to automate the microinjection process for Drosophila melanogaster embryos. This technology can also be adapted for use with various other organisms. This robot eliminates the need for manual microinjection protocols, which are operator-dependent, time-consuming, and

require significant training. The robot uses machine learning (ML) models trained to detect individual embryos on agar plates and guides microinjection needles to perform microinjections at specific locations in each detected embryo. This robot can be operated by individuals with no prior experience and surpass human capabilities

in terms of microinjection speed, performing at a rate six times faster than humans. In AIM 1 of this proposal, we will evaluate the efficacy and generalizability of the Autoinjector technology for transgenesis across a wide spectrum of Drosophila experiments. This involves testing and refining the automated microinjection process for different genetic backgrounds, microinjection locations, microinjectant

compositions, construct sizes, auxiliary plasmids or transgenes, as well as various operational variables such as embryo laying conditions, culture media, DNA concentration, and solution viscosity. AIM 2 of this proposal we will innovate the ML algorithms to address two identified failure modes in the automated

microinjection process. These failure modes are (1) the inability to detect individual embryos when they are clustered together on the embryo collection plates and (2) the clogging of pipettes during automated microinjection. We will develop ML-guided robotic algorithms to mitigate these issues and improve the overall

performance of the system.