Single-molecule localization microscopy error is sensor dependent and larger than theory predicts

Since the advent of stochastic localization microscopy approaches in 2006, the number of studies employing this strategy to investigate the subdiffraction limit features of fluorescently labeled structures in biology, biophysics and solid state samples has increased exponentially. Underpinning all these approaches is the notion that the position of single molecules can be determined to high precision, provided enough photons are collected. The determination of exactly how precisely, has been demanded to formulas that try to approximate the so-called Cramer-Rao lower bound based on input parameters such as the number of photons collected from the molecules, or the size of the camera pixel. These estimates should, however, be matched to the experimental localization precision, which can be easily determined if, instead of looking at single beads, we study the distance between a pair. We revisit here a few key works, observing how these theoretical determinations tend to routinely underestimate the experimental localization precision of the order of a factor 2. A software-independent metric to determine, based on each individual setup, the appropriate value to set on the localization error of individual emitters is provided.