Fast and Accurate Three-Dimensional Point Spread Function Computation for Fluorescence Microscopy

J. Li, F. Xue and T. Blu, J. Opt. Soc. Am. A, 34(6), pp. 1029-1034, 2017.

This set of codes is a fast approximation of 3D PSF model in fluorescernce microscopy, based on Gibson-Lanni model. See the page for details.

Fast approximation Current approximation can be 498 times faster than the state-of-the-art tool PSF Generator for size 511x511x255. Note the PSF Generator uses Java language with mult-threading, while the proposed uses pure Matlab without any compiled routine.

Usage of the Proposed Method

If the parameters of the microscope are not assigned, default setting will be loaded, see MicroscPSF for details.

params.size = [256 256 128];
tic;
PSF = MicroPSF(params);
toc

GUI

Even faster?

The last interpolation step is implementated using mex files, which reduce the time cost 50%.

cd Utilities
mex transformation.cpp

params.size = [256 256 128];
params.fastcom = 1;
tic;
PSF = MicroPSF(params);
toc

System Tested

Update log

• (04 May, 2017) Add mex implementation, the speed is further improved 2 times!
• (28 Dec, 2016) Refined basis functions.
• (09 Oct, 2016) Optimized code, the speed has been improved 10 times faster!

Related

• Matlab code: https://github.com/hijizhou/MicroscPSF-Matlab.

• ImageJ plugin: https://github.com/hijizhou/MicroscPSF-ImageJ.

• Icy plugin: https://github.com/hijizhou/MicroscPSF-Icy.

• Python code (by Dr. Kyle Douglass): http://kmdouglass.github.io/posts/implementing-a-fast-gibson-lanni-psf-solver-in-python.html

Acknowledgement

PSFgenerator

License

Copyright (c) 2017, Jizhou Li, Feng Xue and Thierry Blu

The source code is released under the MIT license.