Novel 3D imaging model for blastocyst evaluation helps predict IVF success

Key takeaways:

• The 3D imaging model achieved more than 90% accuracy in identifying blastocyst features.
• Blastocyst overall volume, cystic cavity volume and surface area were positively correlated with clinical pregnancy rate.

A novel 3D imaging model that identifies blastocyst features may help to improve embryo selection for IVF success, according to findings presented at the European Society of Human Reproduction and Embryology annual meeting.

“This study provides a method for constructing 3D blastocysts using embryo images from daily time-lapse devices,” Bo Huang, MD, an embryologist at the Reproductive Medicine Center of Tongji Hospital, Tongji Medicine College at Huazhong University of Science and Technology in Wuhan, China, told Healio. “This model can provide specific values for important parameters of blastocysts. For example, blastocyst volume, inner cell mass volume, etc.”

Huang and colleagues conducted a study with data from 2,141 frozen-thaw single blastocyst transfer cycles from women younger than 40 years with an endometrial thickness of 7 mm to 16 mm from Tongji Hospital from 2020 to 2021. Using a 3D imaging model, researchers created models of each blastocyst with detailed data on outer layers and inner cell masses. Researchers analyzed the models to find new features and determine how they relate to successful pregnancies with IVF.

Researchers obtained blastocyst trophectoderm epithelium and inner cell mass through deep learning and segmentation networks. Blastocyst texture information was extracted and mapped to the 3D imaging model, which was compared with fluorescence imaging of blastocysts.

The 3D imaging model achieved more than 90% accuracy. Blastocyst overall volume, cystic cavity volume and surface area were positively correlated with clinical pregnancy rate, with differences observed when overall volume was more than 179 wµm³, cystic cavity volume was 165 wµm³ and surface area was 25.8 wµm².

In addition, researchers noted no significant difference in clinical pregnancy rate in terms of inner cell mass. However, researchers observed a higher clinical pregnancy rate ranging from 0.715 to 0.912 for inner cell mass spatial aspect ratio and a rate ranging from 0.92 to 0.943 for inner cell mass surface sphericity.

The number of blastocyst trophectoderm epithelium cells was positively correlated with clinical pregnancy rate and trophectoderm epithelium circumference, area and variance, and aspect ratio and variance were all significantly correlated with clinical pregnancy rate.

“This provides a more specific method for clinical evaluation of blastocysts,” Huang said. “However, moving forward this study requires further improvement of the model’s accuracy and wider validation.”