Structured Illumination Microscopy Based On Digital Micromirror Arrays

Structured Illumination Microscopy(SIM)has the characteristics of low illumination and super-resolution imaging,which is advantageous in dynamic in vivo cell imaging and has become a research hotspot in optical microscopy.The conventional excitation pattern generation method is usually generated by phase grating or spatial light modulator,which has the defects of slow conversion speed,polarization dependence and low streak modulation depth.In this paper,we propose a scheme based on a digital micromirror device(DMD)to generate excitation patterns,and realize a structured illumination imaging system with high-speed streak conversion and high contrast.The paper systematically analyzes the principle of SIM reconstruction imaging,compares the effects of Wiener filtering and Hessian filtering on the reconstruction imaging quality,and carries out modeling and simulation studies.Through the simulation,it is pointed out that SIM reconstructed images have the resolution advantages compared to ordinary images,and for image sources with low signal-to-noise ratios,the Hessian SIM has an advantage over the Wiener SIM in suppressing reconstruction artifacts..In the experiment,The streak light field with high spatial frequency and large modulation depth is realized based on the first-order diffraction streak interference principle of DMD,and the effects of different graphic parameters of DMD on the direction,phase and spatial frequency magnitude of the streaks are analyzed,and a general rule about the streak spatial frequency control is obtained.Based on the standard sample,a line resolution of 330 nm was achieved under 40×objective observation,which was improved by 1.34 times compared with uniform illumination imaging.By experimentally comparing the performance of Wiener SIM and Hessian SIM,it is pointed out that Hessian SIM has the ability to eliminate artifacts while maintaining resolution.The limit resolution of linearly excited SIM can be improved by a factor of 2.73 according to our spatial frequency selection model.