TY - GEN
T1 - Three-Dimensional Reconstruction of Early Embryos Based on Controllable Acoustofluidic Rotation
AU - Bai, Chenhao
AU - Chen, Zhuo
AU - Li, Yunsheng
AU - Liu, Fengyu
AU - Huang, Qiang
AU - Arai, Tatsuo
AU - Liu, Xiaoming
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - High-precision three-dimensional imaging of live embryos is hindered by photobleaching, invasive labels, and the mechanical constraints of conventional rotation methods. To overcome these limitations, we introduce a fully integrated, noncontact acoustofluidic platform for controllable multi-axis embryo rotation and reconstruction. A piezoelectric transducer drives microbubble resonance at the tip of a glass microtube, generating programmable vortical flows that gently and stably spin embryos along both vertical and horizontal axes. We use a deep learning method (Mask R-CNN) to automatically delineate key embryonic structures in real-time. Rotation angles are then estimated via the tracker on surface feature points, and a multiview tomographic algorithm reconstructs the full 3D embryo contour without fluorescent markers. In mouse embryo reconstruction, our system achieves ∼ 3 μ m reconstruction accuracy and reduces imaging time by over 40% compared to manual methods. This acoustofluidic approach opens new avenues for noninvasive, high-throughput embryonic morphokinetic analysis and quality assessment in assisted reproduction and biology.
AB - High-precision three-dimensional imaging of live embryos is hindered by photobleaching, invasive labels, and the mechanical constraints of conventional rotation methods. To overcome these limitations, we introduce a fully integrated, noncontact acoustofluidic platform for controllable multi-axis embryo rotation and reconstruction. A piezoelectric transducer drives microbubble resonance at the tip of a glass microtube, generating programmable vortical flows that gently and stably spin embryos along both vertical and horizontal axes. We use a deep learning method (Mask R-CNN) to automatically delineate key embryonic structures in real-time. Rotation angles are then estimated via the tracker on surface feature points, and a multiview tomographic algorithm reconstructs the full 3D embryo contour without fluorescent markers. In mouse embryo reconstruction, our system achieves ∼ 3 μ m reconstruction accuracy and reduces imaging time by over 40% compared to manual methods. This acoustofluidic approach opens new avenues for noninvasive, high-throughput embryonic morphokinetic analysis and quality assessment in assisted reproduction and biology.
UR - https://www.scopus.com/pages/publications/105030464407
U2 - 10.1109/CBS65871.2025.11267525
DO - 10.1109/CBS65871.2025.11267525
M3 - Conference contribution
AN - SCOPUS:105030464407
T3 - 2025 IEEE International Conference on Cyborg and Bionic Systems, CBS 2025
SP - 212
EP - 217
BT - 2025 IEEE International Conference on Cyborg and Bionic Systems, CBS 2025
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2025 IEEE International Conference on Cyborg and Bionic Systems, CBS 2025
Y2 - 17 October 2025 through 19 October 2025
ER -