photonpy 1.0.23

CUDA-Based image processing for single molecule localization microscopy

A C++/CUDA toolbox with python bindings for doing single molecule microscopy image processing.

Features include:

• Super fast spot detection and extraction on tiff files (~1 KHz on my dell xps laptop)

• Max-Likelihood fitting of various 2D Gaussian models

  (XY with or without fixed sigma, 3D using astigmatism)

• SIMFLUX (https://www.nature.com/articles/s41592-019-0657-7)

• Easy rendering of results: Piccaso Render compatible HDF5 export (https://github.com/jungmannlab/picasso)

• sCMOS-corrected MLE (See https://www.nature.com/articles/nmeth.2488)

• Phasor-based SMLM (https://aip.scitation.org/doi/full/10.1063/1.5005899)

• A C++ templated approach to quickly implement new PSF models as long as the first derivative can be computed.

Work in progress:

• Drift correction using redundant cross-correlation (needs reliability checks)

• Cubic spline PSFs

Currently compiled for CUDA 10.1 / Windows x64 using Visual Studio 2019

Credits to Willem Melching (@pd0wm) for general debugging and implementation of astigmatic Gaussian PSF models, as well as yet unpublished code.