The impulse response of a microscope to a point-like source is described by the point-spread function (PSF). Because a single molecule emitter can be treated as an incoherent point source in SMLM data, the result of fitting a PSF model to an image of a single molecule is an estimate of the molecular position, its imaged size, and its intensity. ThunderSTORM offers a choice between several PSF models and fitting procedures based on (weighted or unweighted) least-squares methods and maximum likelihood estimation [1, 2].

It has been shown [4, 3] that the Gaussian function provides a very good approximation of the real PSF of a microscope. This is mainly due to pixelation effects and the presence of noise, which makes the difference between the Gaussian function and the real PSF negligible. The advantage of Gaussian PSF models are their simplicity, robustness, and computational efficiency.