Artificial Intelligence to Choose Embryos with Greatest Potential
ASRM16 October 2018
Denver, CO- Innovation in embryo grading is taking the form of artificial intelligence systems that can grade embryos and accurately predict ART outcomes. Australian and US groups presented new research Wednesday at ASRM’s Scientific Congress and Expo.
A team in the United States has modeled a convolutional neural network to accurately predict the morphological quality of blastocysts based on time lapse images.
The study utilized 50,392 images of 10,148 blastocysts. The blastocysts were assigned quality grades- good, fair, or poor- based on statistically different implantation outcomes. 18,000 of these images were used to train the Inception IV algorithm, then it was tested using the remaining images.
The algorithm was 97.52% accurate in discriminating between poor and good blastocysts
Meanwhile in Australia, Aengus Tran and his colleagues have developed an artificial intelligence system that 93% of the time correctly predicts that a particular embryo will progress to fetal heartbeat. The fully automated system analyzes time lapse video sequences and requires no human input, and thus is not subject to embryologist variability.
The AI was trained to use time lapse videos sequences to analyze the development of embryos and predict by identifying spatial temporal features, independent of maternal age, whether an embryo would result in a pregnancy with a fetal heartbeat. Eight laboratories in four countries participated from 2014 to early 2018.
The 1,603 patients in the study ranged in age from 22 to 50 with an average age of 35.6, and all embryos (10,208 of them) were included regardless of their stage of development or graded quality.
Amy Sparks, PhD, President-Elect of the Society for Assisted Reproductive Technology, SART, commented, “Artificial Intelligence as applied to analyzing embryo quality and potential holds great promise for improving patients’ chances of achieving a successful pregnancy sooner rather than later.”
O-209 N. Zaninovic et al, Assessing Human Blastocyst Quality Using Artificial Intelligence (AI) Convolutional Neural Network (CNN)
O-265 A. Tran et al, Artificial Intelligence as a Novel Approach for Embryo Selection