O-125 Development of an artificial intelligence embryo witnessing system to accurately track and identify patient specific embryos in a human IVF laboratory. (6th August 2021)

• Record Type: Journal Article
• Authors: Bormann, C; Kanakasabapathy, M; Thirumalaraju, P; Dimitriadis, I; Souter, I; Hammer, K; Shafiee, H
• Is Part Of: Human reproduction. Volume 36:Supplement 1(2021)
• Publication Date: 2021-08-06
• Subjects: Human reproduction -- Periodicals; 618
• DOI: 10.1093/humrep/deab126.050
• Languages: English
• ISSNs: 0268-1161

Abstract

Abstract: Study question: Can convolutional neural networks (CNN) be used as a witnessing system to accurately track and identify patient specific embryos at the cleavage stage of development? Summary answer: We developed the first artificial intelligence driven witnessing system to accurately track cleavage and blastocyst stage embryos in a human ART laboratory. What is known already: There are reports of human errors in embryo tracking that have led to the births of children with different genetic makeup than their birth parents. Clinical practices rely on manual identification, barcodes or radio-frequency identification technology to track embryos. These systems are designed to track culture dishes but are unable to monitor developing embryos within the dish to help ensure an error-free patient match. Previously, we developed an AI witnessing system to track blastocysts with 100% accuracy. The goal of this study was to determine whether an AI witnessing system could be developed that accurately tracks cleavage stage embryos. Study design, size, duration: A pre-developed deep neural network technology was first trained and tested on 4944 embryos images. The algorithm processed embryo images for each patient and produced a unique key that was associated with the patient ID at 60 hpi, which formed our library. When the algorithm evaluated embryos at 64 hpi it generated another key that was matched with the patient's unique key available in the library. Participants/materials, setting, methods: A total of 3068 embryos from 412 patients were examined by the CNN at both 60 hpi and 64 hpi. These timepoints were chosen as they reflect the time our laboratory evaluates Day 3 embryos (60 hpi) and the time we move them to another dish and prepare them for transfer (64 hpi). The patient cohorts ranged from 3-12 embryos per patient. Main results and the role of chance: The accuracy of the CNN in correctly matching the patient identification with the patient embryo cohort was 100% (CI: 99.1% to 100.0%, n = 412). Limitations, reasons for caution: Limitations of this study include that all embryos were imaged under identical conditions and within the same EmbryoScope. Additionally, this study only examined fresh Day 3 embryos cultured over a span of 4 hours. Future studies should include images of fresh and frozen/thawed embryos captured using different imaging systems. Wider implications of the findings: This study describes the first artificial intelligence-based approach for cleavage stage embryo tracking and patient specimen identification in the IVF laboratory. This technology offers a robust witnessing step based on unique morphological features that are specific to each individual embryo. Trial registration number: This work was partially supported by the Brigham Precision Medicine Developmental Award (Brigham Precision Medicine Program, Brigham and Women's Hospital), Partners Innovation Discovery Grant (Partners Healthcare), and R01AI118502, and R01AI138800.