Integrative Sciences and Engineering Programme, NUS Graduate School, National University of Singapore, Singapore.
Department of Biomedical Engineering, National University of Singapore, Singapore.
J R Soc Interface. 2021 Jun;18(179):20210248. doi: 10.1098/rsif.2021.0248. Epub 2021 Jun 16.
Optical flow algorithms have seen poor adoption in the biological community compared with particle image velocimetry for quantifying cellular dynamics because of the lack of proper validation and an intuitive user interface. To address these challenges, we present OpFlowLab, an integrated platform that integrates our motion estimation workflow. Using routines in our workflow, we demonstrate that optical flow algorithms are more accurate than PIV in simulated images of the movement of nuclei. Qualitative assessment with actual nucleus images further supported this finding. Additionally, we show that refinement of the optical flow velocities is possible with a simple object-matching procedure, opening up the possibility of obtaining reasonable velocity estimates under less ideal imaging conditions. To visualize velocity fields, we employ artificial tracers to allow for the drawing of pathlines. Through the adoption of OpFlowLab, we are confident that optical flow algorithms will allow for the exploration of dynamic biological systems in greater accuracy and detail.
与用于量化细胞动力学的粒子图像测速法相比，光流算法在生物界的应用较少，因为缺乏适当的验证和直观的用户界面。为了解决这些挑战，我们提出了 OpFlowLab，这是一个集成平台，集成了我们的运动估计工作流程。使用我们工作流程中的例程，我们证明在细胞核运动的模拟图像中，光流算法比 PIV 更准确。使用实际细胞核图像进行的定性评估进一步支持了这一发现。此外，我们还表明，通过简单的物体匹配过程可以对光流速度进行细化，从而有可能在不太理想的成像条件下获得合理的速度估计。为了可视化速度场，我们采用人工示踪剂来绘制轨迹线。通过采用 OpFlowLab，我们有信心光流算法将能够更准确、更详细地探索动态生物系统。
