Explainable Embryo Grading
in development
Pixel-precise semantic segmentation automatically maps the geometric boundaries of sub-micron cell layers, replacing subjective human visual bias with hard structural data. Every structure the model reads can be isolated on the specimen and checked by eye.
Plate B — Day-5 human blastocyst, brightfield. The four regions are hand-traced boundaries from the annotation set, not model output. Motion paused by system preference.
toggle layers to isolate structures
Inner Cell Mass
The inner cell mass is the cluster of cells that develops into the fetus itself. A well-defined, densely packed ICM is associated with a higher likelihood of implantation.
Select Confidently
Every embryo in the cycle is ranked in one view, each score shown beside the image it was read from. The embryologist sees what the model saw, how sure it is, and keeps the final call.
Inventory
Patient 5
Olivia Brown
?02-12-2026
Key Features
AI Ranking
Every embryo in the cycle is scored and ordered, giving the embryologist one more reference alongside their own read.
Morphology Grading
Expansion stage, inner cell mass and trophectoderm are assessed automatically, so grading holds consistent between embryologists.
Report Generation
Selected embryos summarise into a single sheet, so what the lab decided can be explained plainly to the patient.