Pre-Clinical Validation of an AI-Powered Automated Robotic Intracytoplasmic Sperm Injection System
Hernandez et al., American Society for Reproductive Medicine (ASRM) 2025 Scientific Congress & Expo
Objective
To evaluate the performance and safety of an automated robotic system designed to perform intracytoplasmic sperm injection (ICSI).
Materials and Methods
A fully automated ICSI system was developed using off-the-shelf components, including an inverted microscope, a heated stage, non-contact laser (Lykos DTS, Hamilton-Thorne), PIEZO actuator (PiezoXpert, Eppendorf), a motorized stage and objective turret, microinjectors and digital cameras. A multi-component, in-house trained AI system integrated sperm selection, tracking, immobilization, and oocyte and sperm segmentation, tool positioning, to govern the complete ICSI workflow.
The fully automated ICSI system was evaluated using sperm samples from human donors injected into rabbit oocytes across 59 injections from 4 semen samples. Two routines were assessed: 1) Sperm routine consisting of AI-guided individual sperm tracking and selection, laser-based immobilization, and loading into the needle; and 2) Oocyte injection routine comprising oocyte holding, zona pellucida laser-thinning, sperm positioning inside the needle, oocyte penetration, and sperm deposition inside the oocyte. The technical performance of these steps was evaluated against pre-specified acceptance criteria.
Results
In the Sperm Routine, the system achieved 100% success in contacting the correct droplet. An in-built quality control check ensured laser-immobilized sperm cells were not loaded in the injection needle when the laser shot hit their head instead of tail (0% incidence). Sperm loss inside the needle was limited to 4% (below the pre-specified <5% threshold), and there were no instances of multiple sperm loading (0% vs. <5% threshold). The mean procedure time was 127±60 seconds per routine.
For the Oocyte Injection Routine, egg finding and focusing success rates were 97.6% and 100%, respectively (>95% threshold). Laser-thinning of the zona pellucida, sperm positioning in the injection needle, and needle insertion all achieved 100% success. Sperm deposition success rate was 90%, slightly below the pre-specified >95%. The mean time for the injection routine was 163±70 seconds (above the pre-specified 120s target).
Conclusions
The ICSI automated system demonstrated efficiency and precision in performing critical procedures with human supervision. The system met or exceeded most predefined performance criteria, with minor deviations in sperm deposition success rate and procedure times. These results support the validation of the system as a promising automated solution for performing ICSI procedures in clinical settings, with potential to standardize outcomes and reduce operator intervention.
Impact Statement
This innovation has the potential to transform clinical ICSI practice by standardizing procedures across fertility centers, thereby reducing operator-dependent variability and potentially improving success rates. The work paves the way for further clinical evaluation.
Support
Financial Support: Institutional