Advanced Photonics Nexus, Volume. 2, Issue 4, 046005(2023)

Untrained neural network enhances the resolution of structured illumination microscopy under strong background and noise levels

Yu He1,†... Yunhua Yao1, Yilin He1, Zhengqi Huang1, Dalong Qi1, Chonglei Zhang2, Xiaoshuai Huang3, Kebin Shi4, Pengpeng Ding1, Chengzhi Jin1, Lianzhong Deng1, Zhenrong Sun1, Xiaocong Yuan2,*, and Shian Zhang1,56,* |Show fewer author(s)
Author Affiliations
  • 1East China Normal University, School of Physics and Electronic Science, State Key Laboratory of Precision Spectroscopy, Shanghai, China
  • 2Shenzhen University, Institute of Microscale Optoelectronics, Nanophotonics Research Center, Shenzhen Key Laboratory of Micro-Scale Optical Information Technology, Shenzhen, China
  • 3Peking University, Biomedical Engineering Department, Beijing, China
  • 4Peking University, School of Physics, State Key Laboratory for Mesoscopic Physics and Frontiers Science Center for Nano-optoelectronics, Beijing, China
  • 5East China Normal University, Joint Research Center of Light Manipulation Science and Photonic Integrated Chip of East China Normal University and Shandong Normal University, Shanghai, China
  • 6Shanxi University, Collaborative Innovation Center of Extreme Optics, Taiyuan, China
  • show less
    Figures & Tables(7)
    Flow chart of ADMM-DRE-SIM for realizing the resolution enhancement of SIM.
    Simulation results for the resolution enhancement of SIM by ADMM-DRE-SIM. (a) The ground truth, (b) wide-field, (c) SIM, and (d) ADMM-DRE-SIM images of the simulated structures, associated with the corresponding frequency spectra and intensity distributions along the labeled lines in the insets. The insets are the enlarged views of the selected areas with yellow squares.
    Simulation results for the effects of background and noise levels on the resolution enhancement of SIM by various algorithms. (a) The ground truth, wide-field, SIM, and ADMM-DRE-SIM images of the simulated structures without the background and noise. (b) The SIM, Wiener deconvolution, RLTV deconvolution, Hessian deconvolution, and ADMM-DRE-SIM images with a uniform background combined with Gaussian noise σ=0.1, 0.3, 0.5, and Poisson noise mixed with Gaussian noise σ=0.5.
    Experimental results for the resolution enhancement of SIM by ADMM-DRE-SIM. (a) The wide-field, SIM, and ADMM-DRE-SIM images of tubulins in MEF cells and (c) actins in NIH/3T3 cells, together with the corresponding intensity distributions along the labeled lines in the insets. The FRCs of the SIM and ADMM-DRE-SIM images for (b) tubulins and (d) actins.
    Experimental results for the effects of the strong background and low SNR on the resolution enhancement of SIM by various algorithms. (a) SIM image of tubulins with strong background. Left, the stitched image with the SIM image on the top and the ADMM-DRE-SIM image on the bottom; right, the enlarged SIM image marked with the yellow square and the images processed by RL, RLTV, Hessian, sparse deconvolution, and ADMM-DRE-SIM algorithms, respectively. (b) SIM image of tubulins with low SNR. The image arrangement is the same as (a).
    Comparison of ADMM-DRE-SIM and modified ADMM-DRE-SIM in structural fidelity. (a) SIM image of tubulins; (b) estimated background map; (c) image by directly subtracting the estimated background map from the SIM image. (d)–(g) The SIM, background subtracted SIM, modified ADMM-DRE-SIM, and ADMM-DRE-SIM images for the selected area marked with the yellow square, respectively.
    Experimental results for the resolution enhancement of SIM by ADMM-DRE-SIM with 2× upsampling. (a) The stitched image of CCPs with the SIM image on the left and the ADMM-DRE-SIM image on the right. (b) and (c) The SIM, ADMM-DRE-SIM, and 2× upsampling ADMM-DRE-SIM images for the selected areas in panel (a), associated with the intensity distributions along the labeled lines.
    Tools

    Get Citation

    Copy Citation Text

    Yu He, Yunhua Yao, Yilin He, Zhengqi Huang, Dalong Qi, Chonglei Zhang, Xiaoshuai Huang, Kebin Shi, Pengpeng Ding, Chengzhi Jin, Lianzhong Deng, Zhenrong Sun, Xiaocong Yuan, Shian Zhang, "Untrained neural network enhances the resolution of structured illumination microscopy under strong background and noise levels," Adv. Photon. Nexus 2, 046005 (2023)

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Research Articles

    Received: Feb. 10, 2023

    Accepted: Jun. 14, 2023

    Published Online: Jul. 7, 2023

    The Author Email: Xiaocong Yuan (xcyuan@szu.edu.cn), Shian Zhang (sazhang@phy.ecnu.edu.cn)

    DOI:10.1117/1.APN.2.4.046005

    Topics