O-185 Far away so close: remote ICSI using an AI-powered robot

Abstract

Abstract Study question Can intracytoplasmic sperm injection (ICSI) be performed remotely using a personal computer and an intelligent robotic micromanipulation system? Summary answer An AI-enhanced robotic micromanipulation system was developed that allows successful remote execution of ICSI by experts or non-experts. What is known already ICSI has become the method of choice for fertilization of eggs during IVF, but it remains technically demanding, requiring manual control of highly specialized equipment by expert embryologists. Limited automation of the procedure has been achieved but most steps are still performed manually. Remote performance of ICSI has never been attempted. Study design, size, duration To achieve automated ICSI, we explored individual commands for each step, issued from an offsite computer. Steps included AI-assisted sperm selection, laser immobilization, random oocyte positioning, opening a path in the zona pellucida prior to needle entry, and Piezo-mediated oolemma breakage. The modified protocols were optimized using mouse oocytes and mouse sperm and then tested using the automated system to inject hamster oocytes with human sperm. Finally, the system was evaluated in consenting patients. Participants/materials, setting, methods The system consists of an inverted microscope, a motorized stage, motorized micromanipulators, laser, piezo-electric actuator, and microinjectors, all controlled by proprietary digital and AI-driven software. Development of this system was a collaborative effort across four countries in three laboratories, including an engineering and embryo research facility, an experimental embryology laboratory, and an IVF clinic. The pilot clinical procedures were performed after full IRB review and trial registration. Main results and the role of chance To assess the safety of using a laser on sperm tails for immobilization, manual ICSI was performed on mouse oocytes with mouse sperm immobilized using different laser settings. Once the method was optimized, fertilization (n = 176, 98.8%) and blastocyst formation (80.7%) in the experimental group were similar to those in the controls without laser (n = 173, 99.4% and 80.6%, respectively) (p > 0.05). Subsequently, 90 blastocysts from the experimental group were transferred to pseudopregnant mice resulting in 40.4% live pups. This was comparable to the results with control embryos (n = 90, 47.8%) (p = 0.098). The remotely controlled robotic system was then tested in hamster oocytes, demonstrating survival (n = 102, 94.1%) similar to the manually injected control group (n = 102, 97.1%) (p = 0.3). In the clinical pilot trial in four patients, 22 oocytes were assigned to remote ICSI, achieving survival (95%), fertilization (81%), and blastocyst formation (65%) comparable to manually performed ICSI controls on sibling oocytes. These results are within the benchmark values established in the Vienna consensus. Throughout all experiments, the system allowed users to successfully perform ICSI remotely issuing commands by pressing single buttons on a computer screen from various locations, ranging in distance from a few meters to over 9,000 kilometers away from the laboratory. Limitations, reasons for caution This experimental ICSI system requires operators in the laboratory to insert and position the microtools and handle the eggs and sperm before and after ICSI. The system also requires a broadband internet connection with at least 50MBPs speed in upload and download. Latency or delayed reaction is a rare possibility. Wider implications of the findings The functionality of a remotely controlled automated ICSI system was validated. The system demonstrates efficiency gains, even with remote operation. The procedures can be programmed to achieve full microscopic ICSI using a single command. This system represents a significant advancement in automation of IVF. Trial registration number Clinicaltrials.gov: NCT06074835 IRB approval: RA-2023-01 IRB ID: CONBIOÉTICA-09-CEI-001-20170131