Abstract Study question Automated sperm selection or human eye-based sperm selection, does it make a difference on IVF lab outcomes and clinical outcomes? Summary answer Single sperm selection by the AI SiD was as proficient as highly experienced embryologists and embryo development may have improved for inexperienced embryologists What is known already The implementation of artificial intelligence in reproductive medicine has proven effective for a variety of applications. One application for this technology is single sperm selection during ICSI. The computer assisted program SiD, was designed to grade sperm, in real-time, based on motility parameters transforming qualitative sperm assessments into a quantitative score. SiD’s assistance can be used to select the most optimal sperm for ICSI. The application of sperm selection by AI is expected to bypass user bias, fatigue, limited training, while also optimising time. However, there are limited studies correlating the usage of AI for sperm selection with ICSI outcomes. Study design, size, duration 1 year period 608 sibling MII oocytes were randomly divided in two groups; 1-ICSI group (n = 301): ICSI performed with sperm selected by the embryologist and 2- SID-ICSI group (n = 307): ICSI performed with sperm selected using SiD software. Priority was given to sperm classified as “Best” by SID software for ICSI.. Morphology selection was performed before ICSI in both groups. Embryos were cultured during 6 days in the same conditions. Participants/materials, setting, methods Sperm preparation was performed using a microfluidic chamber (Zymot™). Sperm selection was performed as previously described, and ICSI was performed following routine procedures. Biological and clinical outcomes were analysed in both groups and compared including fertilization rate, cleavage rate, embryo and blastocyst development rate, top quality embryo rate, blastocyst development rate and pregnancy rates. Biological outcomes in both groups were also compared based on embryologist seniority. Embryo morphokinetics parameters were analysed using time-lapse system. Main results and the role of chance There was a non significant trend towards better outcomes in the SID-ICSI group for all biological outcomes including fertilization rate (83,1% vs 82,4%), cleavage rate (97,6% vs 97,2%), day 3 embryo development rate (72,9% vs 70,6%), blastocyst development rate at day 5 (49,0% vs 44,8 %), good quality blastocyst development rate at day 5 (45,1% vs 41,5%), and top quality blastocyst development rate at day 5 (25,9% vs 22,2%). Similarly, a non-significant increase was observed in all these outcomes when sperm selection is performed by a junior embryologist. Embryo development was monitored using a timelapse system. Our data show that 2nd polar body extrusion (tPB2) (4,0h vs 4,8 h [p = 0,005]) and early cleavage (t2) (27,3h vs 28,7h [p = 0,05]) happen significantly earlier when SiD is used for ICSI. No significant difference was observed for all other fertilization events (cytoplasmic wave,tPN1,tPN2,presence of cytoplasmic halo,tPNf,and cytoplasmic halo disappearance), and other cleavage timings (t3->t10,tM,tSC,tEC,tSB), and embryonic cell cycles (ECC1,ECC2,ECC3,s2,s3). Euploid rate obtained with in the 2 groups was comparable (53,3% in SID-ICSI vs 44,4% in ICSI). Although there was a trend toward higher cumulative early and clinical pregnancy rate after ICSI-SID, the difference did not reach significance (43,1%vs34,0% and 27,5%vs21,2%, respectively). Limitations, reasons for caution The ZymotTM microfluidic chamber was used for sperm preparation. This device optimises sperm selection, giving rise to an already optimised population of spermatozoa. Future investigations evaluating the effect of SiD automated sperm selection software on biological outcomes in samples prepared with a less effective technique would potentially yield conversing results. Wider implications of the findings This pilot study demonstrated that the usage of the automated sperm selection software SiD gives rise to similar biological outcomes and embryo morphokinetics, indicating its effectiveness in sperm selection. We expect this program would be useful in the presence of less experienced laboratory members helping by standardizing sperm selection. Trial registration number NA
Read full abstract