Robust and highly performant ring detection algorithm for 3d particle tracking using 2d microscope imaging.
Robust and highly performant ring detection algorithm for 3d particle tracking using 2d microscope imaging.
复制标题
DOI：
10.1038/srep13584
复制
发表时间：
2015-09-02
影响因子：
4.6
通讯作者：
Afik E
中科院分区：
文献类型：
--
作者：
Afik E
三维粒子跟踪是研究显微镜下动力学的重要工具，即微流控装置中的流体动力学、细菌趋向性、细胞运输。三维位置可以通过单独使用二维成像来确定，通过测量由失焦荧光粒子产生的衍射环，在单个相机上成像。在这里，我提出了一种环检测算法，它具有很高的检测率，对环遮挡、内含物和重叠所带来的挑战具有鲁棒性，并且即使在彼此靠近时也可以分辨粒子。由于其高性能和低内存占用，它能够进行实时分析。该算法是圆形霍夫变换的后代，通过解决分类问题，解决了当粒子数量不一定固定时，有效跟踪多个粒子轨迹的需要，并克服了在复杂参数空间中由环形簇和噪声引起的局部最大值的挑战。这里介绍的几个算法概念在其他情况下可能是有利的，特别是在处理嘈杂和稀疏数据时。该实现仅基于开源和跨平台软件包，因此易于分发和修改。在70 Hz的微流控实验中实现了实时多粒子跟踪，检测率超过94%，误检率仅为1%。
Three-dimensional particle tracking is an essential tool in studying dynamics under the microscope, namely, fluid dynamics in microfluidic devices, bacteria taxis, cellular trafficking. The 3d position can be determined using 2d imaging alone by measuring the diffraction rings generated by an out-of-focus fluorescent particle, imaged on a single camera. Here I present a ring detection algorithm exhibiting a high detection rate, which is robust to the challenges arising from ring occlusion, inclusions and overlaps, and allows resolving particles even when near to each other. It is capable of real time analysis thanks to its high performance and low memory footprint. The proposed algorithm, an offspring of the circle Hough transform, addresses the need to efficiently trace the trajectories of many particles concurrently, when their number in not necessarily fixed, by solving a classification problem, and overcomes the challenges of finding local maxima in the complex parameter space which results from ring clusters and noise. Several algorithmic concepts introduced here can be advantageous in other cases, particularly when dealing with noisy and sparse data. The implementation is based on open-source and cross-platform software packages only, making it easy to distribute and modify. It is implemented in a microfluidic experiment allowing real-time multi-particle tracking at 70 Hz, achieving a detection rate which exceeds 94% and only 1% false-detection.