Fig. 1. Imaging principle of SFF. (a) Schematic of SFF system ^[25]; (b) lens imaging model when object is moving^[26]

Fig. 2. Imaging results of SFF. (a) Image sequence of sunken plane with NSST^[42]; (b) image sequence of coins^[42]; (c)-(e) comparison of SFF results on TFT-LCD color filter^[48]

Fig. 3. Schematics of SIM system based on DMD projection and LED lighting^[65]. (a) Schematic of TIR prism; (b) structure of system, in which ① is LED light guide, ② is DMD, ③ is sample stage, and ④ is CCD camera

Fig. 4. Optical sectioning SIM imaging results and imaging devices. (a) 3D imaging results of lunar meteorite^[74]; (b) 3D imaging results of amber^[73]; (c) schematic of imaging system^[73]; (d) physical picture of imaging device^[73]

Fig. 5. Imaging results of azurite^[76]. (a)-(c) Three-view maximum projection images of azurite; (d) 3D height map of Fig. 5 (a) and height fluctuation distribution on black line segment; (e) detail image on malachite; (f) image of malachite restored by HSV decoding algorithm; (g) detail image on white crystal; (h) image of white crystal restored by HSV decoding algorithm

Fig. 6. GANs for image reconstruction in microscopy^[88]

Fig. 7. 3D imaging of caenorhabditis elegans head neuron nuclei obtained using Deep-Z^[15]

Fig. 8. Volumetric imaging based on Recurrent-MZ^[16]. (a) Recurrent-MZ volumetric imaging framework; (b) Recurrent-MZ network structure

Fig. 9. Overview of WFCON-Net architecture^[89]. (a) WFCON-Net based on 3D GAN; (b) network architecture of generator;

Fig. 10. Image processing results of deep learning model. (a)-(d) Typical coronal maximum intensity projections of four mouse brains^[90]; (e) application of cGAN model in radical prostatectomy^[91]

Fig. 11. Schematics of FC-WFM-Deep imaging system^[92]. (a) Training network using wide field (WF) image and corresponding composite optical sectioning (OS) reference image; (b) reconstructing high-quality output image consisting of OS in DOF using trained network