Fig. 4: The reconstruction results of PAR-SIM.

a 100 nm diameter fluorospheres with 3-angle-3-phase illumination under 0.2 ms pattern display exhibits exceeded 110 nm resolution, and the 7 fluorospheres squared with dashed lines are selected to do resolution statistics in e, while the red one is selected to display in c, d for comparing resolution between wild field and PAR-SIM. b 100 nm diameter fluorospheres with 2-angle-3-phase illumination under 0.4 ms pattern display exhibits perfect 100 nm resolution, and the 7 fluorospheres squared with dashed lines are selected to do resolution statistics in e, while the red one is selected to display in c, d for comparing resolution between wild field and PAR-SIM. The resolution difference comes from various exposure time, thus SNR. 2-angle-3-phase and 3-angle-3-phase modes are used to demonstrate the successful reconstruction of PAR-SIM regardless the illumination strategies, and these two exposure durations of SLM are used to test reconstructed resolutions. c The red dashed square selects the fluorosphere exposed in 0.2 ms and 0.4 ms pattern display mode to exhibit the resolution enhancement. d The profile along white lines in c shows PAR-SIM and wild field resolution are 113 nm and 275 nm, respectively, under 0.2 ms mode. PAR-SIM and wild field resolution are 99 nm and 258 nm under 0.4 ms mode due to the higher SNR. e The 7 selected spheres statistics to resolution in these two exposure modes. There are two 0.2 ms reconstructed sub-ROIs and two 0.4 ms reconstructed sub-ROIs collected to calculate the resolution. And 0.4 ms mode shows a more perfect resolution on 100 nm and 0.2 ms mode exhibits exceeding 110 nm resolution. The PAR-SIM successfully reconstruct (f) the actin filaments, which are exposed with 488 nm 0.4 ms fringe in 2-Angle-3-Phase illumination. Due to the nonexistent reconstruction in 2-Angle illumination for HiFi-SIM, only PAR-SIM, HessianSIM, and fairSIM are compared. PAR-SIM has extraordinary reconstruction, while the other three only show the noise and artifacts. Using the 3-Angle-3-Phase and 0.4 ms exposure, g the microtubule structures in 561 nm excitation have been reconstructed by all algorithms, however, only PAR-SIM exceedingly eliminates the artifacts and the enlarged subset in white dashed square is its result compared with the HiFi-SIM and fairSIM. Meanwhile, HessianSIM has a slightly worse reconstruction in this area (not shown). The actin filament sample (h) excited in 488 nm uses the same illumination strategy, but all three HessianSIM, fariSIM, and HiFi-SIM fail to reconstruct and are buried into severe artifacts. All the conditions are same in the mitochondrial membrane (i) excited in 561 nm, only PAR-SIM works and exceeds the other three candidates with such low signal raw data. From f–i the structure along white arrows are profiled in WF (light blue lines) vs PAR-SIM (light pink lines), and PAR-SIM distinguish these configurations successfully. Scale bars: f–i 500 nm, c 100 nm. Meanwhile, Table 4 lists all the raw data’s averaged SNR calculation from each figure