Phys. Rev. X 7, 041007 (2017)

Although microscope technology has progressed steadily, the ability to extract quantitative information from microscopy images has lagged behind. Matthew Bierbaum and colleagues from Cornell University, USA, now report a generic, methodological approach to extract all the useful information theoretically contained in a complex microscope image. The method, called parameter extraction from reconstructing images (PERI), creates an optical model of a microscope based on the physics of the light interacting with the sample and with the microscope’s optical train. In the process, the fluorescent dye is distributed unevenly throughout the sample, the dyed sample is illuminated unevenly by the laser, and the resultant image is noisy and blurred due to diffraction. Least-squares fitting is then carried out for every parameter in the model to find the correct particle positions, particle radii, illumination field and point-spread function. It takes between 1 and 24 hours for a large confocal image to be analysed, depending on previous knowledge of the microscope’s global parameters. The team demonstrates this approach with a confocal image of colloidal spheres and report measurements of particle positions and radii to within 3 nm accuracy, a 10–100 times improvement over existing methods. The open-source code is available online for researchers to analyse their existing images.