Structured illumination microscopy using unknown speckle patterns

Using spatially non-uniform illumination significantly improves the resolution of light microscopy1. Indeed, frequency mixing between the object and the illumination permits the recovery of object frequencies beyond the diffraction-limited detection band pass2,3,4,5. However, the image reconstruction process requires a precise knowledge of the illumination patterns (usually focused or periodic) and therefore sophisticated stable mountings6,7. Here, we show, both theoretically and experimentally, that image reconstruction can be performed without knowing the illumination patterns, provided that their average is roughly homogeneous over the sample. Using blind structured illumination microscopy (blind-SIM), a resolution about two times better than that of conventional wide-field microscopy is obtained by simply illuminating the sample with several uncontrolled random speckles. Our approach is insensitive to specimen or aberration-induced illumination deformations, does not require any calibration step or stringent control of the illumination, and dramatically simplifies the experimental set-up. By illuminating a sample with several uncontrolled random speckles and implementing a blind structured illumination microscopy algorithm, researchers demonstrate that image reconstruction can be achieved without knowing the original illumination pattern, at a resolution two times better than that of conventional wide-field microscopy.