JBRA Assisted Reproduction 2025;29(Suppl.2 SBRA 2025):189
Poster Presentation
29^th Annual Congress of the SBRA. São Paulo/SP - Brazil, 2025
doi: 10.5935/1518-0557.20263896

P-177. Predicting Euploidy Rates through AI-Driven Oocyte Quality Assessment with Magenta Software

Karla Pacheco de Melo¹, Maithê Alves Fanasca Fiocco¹, Michelli Suemi Tanada¹, Ivan Henrique Yoshida¹, Emerson Barchi Cordts¹, Caio Parente Barbosa¹

¹ Instituto Ideia Fértil de Saúde Reprodutiva - São Paulo - SP - Brasil

Objective: Evaluate whether Magenta, an artificial intelligence–based imaging tool that assigns predictive morphological scores to oocytes, correlates with euploidy rate in assisted reproductive treatments.
Methods: This retrospective cohort study included 83 patients undergoing ICSI between February 2023 and August 2024, with a total of 738 oocytes evaluated. Oocytes from cycles with male factor infertility (oligozoospermia or testicular/epididymal sperm) were excluded. Post-ICSI images were captured via Embryoscope Plus^® and analyzed in Magenta. Oocytes received scores (0–10) and were categorized into four groups (G1: 0–2.5; G2: 2.6–5.0; G3: 5.1–7.5; G4: 7.6–10.0). Correlation between Magenta scores and euploidy rates was assessed using Spearman's test.
Results: A previous study conducted by our team revealed a positive correlation between Magenta scores and blastocyst rate (rs=0.270, p<0.001), indicating a statistically significant progressive increase in blastocyst rates from G1 (33.7%) to G2 (54.6%), G3 (62.8%), and G4 (68.2%). Analysis of euploidy rates in relation to the Magenta score revealed a positive correlation (rs=0.094, p=0.045). Despite a statistically significant result, the correlation coefficient close to zero indicates that other variables are likely to contribute to the chromosomal normality of embryos.
Conclusion: Magenta provides an objective, artificial intelligence–based tool that complements conventional oocyte assessment in IVF laboratories, assisting embryologists in identifying oocytes with a higher probability of generating euploid embryos. Nevertheless, given the weak correlation observed, its use cannot supplant standard morphological and clinical evaluations. Additional prospective studies are warranted to improve its predictive performance and to validate its clinical applicability across diverse populations.