Assessment of Human Blastocyst using Deep Learning Algorithm

Human blastocyst is an embryo on its 5th day of development. The formation of 32 cell stage is called Blastocyst stage and its size is about 0.2mm. Blastocyst analysis is to automate blastocyst morphology by analyzing with multiple images. A fertilized egg is cultured for five days before being put into the uterus when using blastocysts in in-vitro fertilization. It might be a more successful fertility treatment alternative than standard in-vitro fertilization. The Blastocyst assessment aims to increase in-vitro fertilization success rates based on women age. Deep learning is an enabling technology to fulfill all of the above requirements and this model helps in assessing the morphology and cellular composition of blastocysts. Approximately 40% of human blastocysts are genetically normal, however this number drops to 25% if the woman was aged over 40 when her eggs were collected. The model performance is evaluated based on accuracy, loss, Precision and recall values. The Higher accuracy in blastocyst assessment can be achieved by training a DenseNet model on a large dataset of elucidated blastocyst images. This Model achieved a significantly higher accuracy of 92% by assessing the blastocyst development based on women age.