We present a number of methods to locate, rapidly and accurately, fluorescent particles imaged with fluorescence and confocal microscopy.
We have implemented the traditional Crocker and Grier methodology to locate particles, with accelerated CPU and GPU libraries. As a result, the particle detection is sufficiently fast and accurate to be fed back into the mechanical control of a microscope stage, enabling real-time target-locking of colloidal clusters, as shown in the video. Because we can follow a growing colloidal cluster as it moves out of the original fixed field of view, we can extend significantly the observation time of these processes, beyond the limits that diffusion typically imposes. We have also been able to observe actively-transported quantum dots (QDs) endocytosed into live cells free to move in three dimensions, for several hours.
We have also looked carefully at all of the stages of these algorithms, and show that accurately locating particles in 3D requires circular convolution kernels, not the square kernels chosen in earlier implementations for performance reasons. While seemingly a trivial detail, in fact using non-circular kernels can lead to the artifactual detection of crystallinity in dense, disordered particle suspensions that do not in reality have significant crystalline order.
- Peter J. Lu, Peter A. Sims, Hidekazu Oki, James B. Macarthur, David A. Weitz, “Target-locking acquisition with real-time confocal (TARC) microscopy” Optics Express. 15, 8702–8712 (2007). [pdf]
- Peter J. Lu, Maor Shutman, Eli Sloutskin, Alexander V. Butenko, “Locating particles accurately in microscope images requires image-processing kernels to be rotationally symmetric” Optics Express. 21, 30755–30763 (2013). [pdf]
- Peter J. Lu, thesis, “Gelation and Phase Separation of Attractive Colloids” Harvard University (2008). [pdf]
- Peter J. Lu, “Confocal Scanning Optical Microscopy and Nanotechnology” in Handbook of Microscopy for Nanotechnology, N. Yao, Z. Wang, Eds. (Kluwer, New York, 2005)pp. 3–24. [pdf]