Non-Invasive Bovine Embryo Assessment using AI and Impedance Spectroscopy

Project Summary/AbstractIn vitro fertilization (IVF) and embryo transfer (ET) are well-established assisted reproductivetechniques in the cattle industry. The protocols have matured through decades of improvementsyet the conception rate following IVF/ET in cattle is around 35-45%. The major limiting factorsfor ET and IVF are labor-intensive protocols dependence on practitioner expertise poor superovulatory response and inaccurate embryo assessment. This research effort aims to solve two ofthe significant challenges mentioned above by (1) improving the labor-intensive embryo dilutionprocess using microfluidics and (2) developing a non-invasive and objective embryo assessmenttechnique. Hawai'i Innovation Laboratory (HIL) in collaboration with Agrimark GeneticsWhitaker Embryo Transfer Services and Hawai'i Pacific University plans to develop a novelembryo assessment technique that is non-invasive and objective unlike the subjective IETS- approved morphological grading system. The proposed technique can assess the embryo viabilityquantifiable by measuring the electrical impedance and observation of morphokinetics such asblastocyst expansion rate blastomere symmetry and percentage of the intact embryonic cellsrepresented by the extruded material in the perivitelline space. The embryo assessment takes placein a microfluidic device that can wash the embryo from the flushing medium easing the embryodilution process. Also the microfluidic chip allows positioning a single bovine embryo betweentwo micro-electrodes under optical microscopy recording time-lapse images analyzingmorphokinetics data and measuring electric impedance circumventing the background noiseconductive media. In prior work HIL has measured the electrical impedance of single Artemiacysts at various phases of development in a microfluidic environment. A distinct pattern ofimpedance changes was observed at different stages of cyst development. The measuredimpedance changes corresponded to physiological changes as the cyst developed. The change inimpedance during the first stage of development provided sufficient quantitative data to predict ifthe cyst would hatch (Rahman et al. IEEE NANOMED 2018 and Matthews et al. IEEE NEMS2017). Also in a recent work the proposer experimented the morphokinetics observations throughtime-lapse image segmentation using Deep Neural Network (DNN) which correlates with itsgenetic health (Huang et al. Reprod Biomed 2019 and Huang st al. Fertil Steril. 2019). Wedemonstrated genetically normal embryos expand more rapidly than genetically abnormalembryos. Building on the above research HIL combines DNN-assisted morphokinetics andImpedance spectroscopy and proposes developing a non-invasive quantitative embryoassessment technique that can determine viability by measuring internal cell information such asmembrane capacitance cytoplasmic conductivity along with the morphokinetic data. During PhaseI of the project the team will design and fabricate a microfluidic device designed for impedancemeasurements on bovine embryos. Also impedance spectra of the embryo will be measured andfitted to the equivalent circuit through non-linear regression to calculate the membrane capacitanceand cytoplasm conductivity. In addition the embryo metabolites will be determined by measuringthe depletion of glucose and pyruvate and the production of lactate to compare the impedance datato their biological health. This effort will result in a commercial tool for theET/IVF practitionerand embryo producers if successful. The tool will help to determine bovine embryo viabilityquantitatively prior to the freezing step. Also the tool can potentially improve the pregnancy rateby 10-15% per transfer by reducing misjudging and discarding of competent embryos and reducingtransfer or less competent embryos. This work will help the NIFA achieve its mission of fosteringan i