The Technology
Over 70 million couples worldwide suffer from infertility while the number of women that undergo assisted reproduction technology (ART) cycles has doubled over the last 15 years. Although much effort is dedicated to improving ART success rates, only 30%-40% of the ART cycles result in pregnancy. Artificial intelligence (AI) is one of the promising technologies for advancing biomedicine. In particular, Deep Convolutional Networks have proven useful in analyzing image-based data such as radiology and pathology.While the in-vivo development of the preimplantation mammalian embryo is dynamic, in clinics all embryos develop under uniform conditions that do not reflect their heterogeneity and maternal clinical history. To date, the decision of which embryo to use for implantation relies on subjective measures that may differ between clinics. Even when objective models are used, reliable grading requires intensive manual labor to extract the embryo features. The dynamic nature of the development process, and its inherent heterogeneity and noise, pose a challenge for using standard image processing to standardize and automate embryo phenotyping. Our goal is to translate our research on embryo development in multiple environments and expertise in AI to a deep learning system that will be able to infer the embryo state and estimate its viability. Moreover, we develop an AI-based apparatus for personalized modulation of the embryo environment to accommodate the single embryo needs. This system will enable rapid screening of the effect of drugs on the developing embryo and will provide a decision support system to assist physicians in embryo selection.
Advantages
- Time saving
- Consistency
- Novel and innovative approach
Applications and Opportunities
- Assisted reproduction technology
- Digital pathology
Business Development Contacts