Chinese Journal of Lasers, Volume. 48, Issue 17, 1707001(2021)

Reconstruction for Cherenkov-Excited Luminescence Scanned Tomography Based on Unet Network

Wenqian Zhang1,2, Jinchao Feng1,2、*, Zhe Li1,2, Zhonghua Sun1,2, and Kebin Jia1,2
Author Affiliations
  • 1Beijing Key Laboratory of Computational Intelligence and Intelligent System, Faculty of Information Technology, Beijing University of Technology, Beijing 100124, China
  • 2Beijing Laboratory of Advanced Information Networks, Beijing 100124, China
  • show less
    Figures & Tables(15)
    A schematic representation of the proposed algorithm
    Phantom and experimental setting used in this study
    Representative experimental data used for network training. (a) Typical examples of true images; (b) typical examples of input images of the Unet network
    Loss curves for training and validation datasets during the training stage. (a) Loss curves for single target dataset; (b) loss curves for two targets dataset
    Reconstructed images by different algorithms at different depths, where the depth of the fluorescent target in (a)--(e) increases from 10 mm to 50 mm with the step size of 10 mm
    Profiles of quantum yield of single target reconstructed image along horizontal direction of the center of the fluorescence target, where the depth of the fluorescent target in (a)--(e) increases from 10 mm to 50 mm with the step size of 10 mm
    Quantitative results for different algorithms. (a) MSE; (b) PSNR; (c) SSIM
    Statistic results for different algorithms. (a) MSE; (b) PSNR; (c) SSIM
    Reconstructed images with different algorithms in case of two targets, where the edge-to-edge distance of two fluorescent targets in (a)--(f) decreases from 5 mm to 0 mm with step size of 1 mm
    Profiles of quantum yield of multiple targets reconstructed image along horizontal direction of the center of the fluorescence target, where the edge-to-edge distance of two fluorescent targets in (a)--(f) decreases from 5 mm to 0 mm with the step size of 1 mm
    Reconstructed images of multiple targets by our algorithm at varied depths
    Reconstructed images of single target at different contrasts. (a) Contrast is 4∶1; (b) contrast is 3.5∶1; (c) contrast is 3∶1; (d) contrast is 2.5∶1; (e) contrast is 2∶1
    Reconstructed images of two targets using the model trained on a single target datasets. (a) Edge-to-edge distance is 5 mm; (b) edge-to-edge distance is 4 mm; (c) edge-to-edge distance is 3 mm; (d) edge-to-edge distance is 2 mm; (e) edge-to-edge distance is 1 mm; (f) edge-to-edge distance is 0 mm
    • Table 1. Comparison of reconstruction time for different algorithms

      View table

      Table 1. Comparison of reconstruction time for different algorithms

      Depth /mmReconstruction time /s
      TikhonovGraph-TVAMPUnetOurs
      1061.35112.11114.2711.2113.06
      2055.47137.93117.7110.5610.43
      3053.27149.97121.0910.2710.08
      4046.80133.25106.319.118.21
      5059.09139.13124.4913.0212.47
    • Table 2. Statistic results for different algorithms

      View table

      Table 2. Statistic results for different algorithms

      AlgorithmMSEPSNR /dBSSIM
      Tikhonov1.40×10-318.1350.383
      Graph-TV7.59×10-422.0950.794
      AMP8.18×10-421.1410.603
      Unet1.04×10-430.4220.897
      Ours6.15×10-532.8210.933
    Tools

    Get Citation

    Copy Citation Text

    Wenqian Zhang, Jinchao Feng, Zhe Li, Zhonghua Sun, Kebin Jia. Reconstruction for Cherenkov-Excited Luminescence Scanned Tomography Based on Unet Network[J]. Chinese Journal of Lasers, 2021, 48(17): 1707001

    Download Citation

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

    Category: biomedical photonics and laser medicine

    Received: Jan. 11, 2021

    Accepted: Mar. 11, 2021

    Published Online: Sep. 1, 2021

    The Author Email: Feng Jinchao (fengjc@bjut.edu.cn)

    DOI:10.3788/CJL202148.1707001

    Topics