Introducing time-lapse for flexible embryo assessment in in vitro fertilization: implications for grading confidence and workflow efficiency

Abstract

Objective

To compare embryologists’ confidence levels and grading concordance when assessing embryo morphology using time-lapse images vs. traditional direct observation under a microscope.

Design

Prospective study.

Subjects

Six embryologists on staff at the laboratory who ranged from junior to senior skillset levels.

Exposure

None.

Main Outcome Measures

Embryologists’ confidence levels in morphology assessments, intraobserver agreement for embryo grading, and assessment duration when using time-lapse images compared with direct observation under microscopes.

Results

From 6,435 morphology assessments across 714 embryos (62 cohorts), high-confidence levels were observed on day 5 for gradings using either direct observation (97.7%) or time-lapse images (94.6%), with no statistically significant difference. Confidence was consistently high (>85%) across days 1, 3, 6, and 7, with direct observation having higher confidence rate than time-lapse-based assessment on days 1 and 3. Intraobserver agreement between direct observation vs. time-lapse-based assessment showed good-to-excellent concordance (κ ≥ 0.60) except for blastocyst morphology grade (κ = 0.58). The highest agreement was observed for day 5–7 embryo disposition decisions, with 94.1% concordance (κ = 0.89). Assessment durations were comparable between the two grading methods.

Conclusion

Embryologists maintained high confidence in morphology assessments using time-lapse systems, even under time constraints. Intraobserver agreement for embryo grading, assessed by comparing the same embryologist’s gradings using time-lapse images vs. direct observation, was comparable to previously reported levels of agreement for traditional morphology gradings. These findings highlight the feasibility of flexible or remote assessments without disrupting embryo culture, enhancing workflow adaptability, and paving the way for advancements with artificial intelligence integration.