OBJECTIVE:

To address inter-observer variability in Gardner embryo grading and investigate whether a newly defined set of morphological indicators can improve grading consensus and better predict clinical outcomes defined by fetal heart tone (FHT).

MATERIALS AND METHODS:

180 static day-5 embryo images were collected from three IVF clinics between 2015 and 2021. Three embryologists with over 5 years of experience independently assigned Gardner grades [INITIAL]. Consensus, defined as unanimous agreement among graders, was achieved in 53 cases (29.4%). A

morphology-based grading index composed of 10 features: 5 representing inner cell mass (ICM) and 5 representing trophectoderm (TE) characteristics. Each image was labeled accordingly, and final consensus grades [FINAL] were assigned. Ordinal logistic regression identified morphological factors for ICM and TE grades, and logistic regression evaluated FHT prediction. Model fit was assessed using Pseudo R².

RESULTS:

The index-based model showed greater explanatory power (Pseudo R² = 28.9%) for FHT compared to the FINAL Gardner model (10.4%), despite both showing borderline significance (p = 0.054 vs. p = 0.073). This limited statistical significance may be attributed to the relatively small sample size (n = 92) of FHT-matched images available for analysis. As shown in Table 1, the associations between the index features and assigned Gardner grades align with findings from previous studies.

CONCLUSIONS:

Our morphology-based index, derived from expert consensus, aligns closely with final grading decisions and demonstrates stronger clinical explanatory power for FHT. The index may offer a more robust and reproducible framework for embryo evaluation than the conventional Gardner system.

IMPACT STATEMENT:

The proposed grading system may reduce subjectivity, enhance expert agreement, and support the development of AI models trained on consensus-based rather than individual assessments.