Pixel Quantum Efficiency Differences and Variance Stabilization for sCMOS Single Molecule Localization Microscopy Data Analysis

ABSTRACT Optimal analysis of single molecule localization microscopy (SMLM) data acquired with a CMOS camera requires compensation for single pixel differences in gain, offset and readout noise. For some CMOS cameras we found that it is also necessary to compensate for pixel differences in sensitivity or relative quantum efficiency (RQE). We present the modifications to the original sCMOS analysis algorithm necessary to correct for these RQE differences. We also discuss the use of the Anscombe transform (AT) for variance stabilization. Removing the variance dependence on the mean allows simpler least squares fitting approaches to achieve the Cramer-Rao bound on the mixed Poisson and Gaussian distributed data typically acquired with an sCMOS camera.