Acta Optica Sinica, Volume. 43, Issue 14, 1434001(2023)

Dual Plug and Play Multi-Energy Computed Tomography Reconstruction Algorithm

Yizhong Wang, Ailong Cai, Ningning Liang, Xiaohuan Yu, Junru Ren, Lei Li, and Bin Yan*
Author Affiliations
  • Henan Key Laboratory of Imaging and Intelligent Processing, Information Engineering University, Zhengzhou 450001, Henan, China
  • show less
    Figures & Tables(15)
    Solving steps of a weighted kernel norm minimization model
    Workflow of the proposed algorithm
    Energy spectrum and eight energy channels used in the experiment
    Truth images of different energy channels. (a) Truth image of the channel [16-22) keV; (b) truth image of the channel [28-31) keV; (c) truth image of the channel [41-50) keV
    Reconstruction results of different algorithms under the incomplete dataset problem. (a) Reconstruction results under the channel [16-22) keV; (b) reconstruction results under the channel [28-31) keV; (c) reconstruction results under the channel [41-50) keV
    ROI comparison results of reconstructed images. (a) Reconstruction results under the channel [16-22) keV; (b) reconstruction results under the channel [28-31) keV; (c) reconstruction results under the channel [41-50) keV
    Pixel values of reconstruction results of different algorithms under incomplete dataset. (a) Reconstruction results under the channel [16-22) keV; (b) reconstruction results under the channel [28-31) keV; (c) reconstruction results under the channel [41-50) keV
    Reconstruction results of different algorithms under the noise interference problem. (a) Reconstruction results under the channel [16-22) keV; (b) reconstruction results under the channel [28-31) keV; (c) reconstruction results under the channel [41-50) keV
    Pixel values of reconstruction results of different algorithms under noise interference. (a) Reconstruction results under the channel [16-22) keV; (b) reconstruction results under the channel [28-31) keV; (c) reconstruction results under the channel [41-50) keV
    Performance of different algorithms after adding different intensity random noise to different sampling data (channel 1)
    Convergence line of the logarithm with RMSE at different problems. (a) Incomplete dataset problem; (b) noise interference problem
    • Table 1. Scanning and reconstruction parameters in the simulation experiment

      View table

      Table 1. Scanning and reconstruction parameters in the simulation experiment

      ParameterValue
      Distance from ray source to detector /mm180
      Distance from ray source to object /mm132
      Number of detector bins512
      Detector bin size /mm0.1
      Size of the reconstructed image256×256
      Pixel size /mm0.15
      Scanning range/(°)60
    • Table 2. Quantitative results of reconstructed images obtained by different algorithms on the simulated data under incomplete dataset

      View table

      Table 2. Quantitative results of reconstructed images obtained by different algorithms on the simulated data under incomplete dataset

      ChannelParameterSARTASD-POCSSISTERProposed algorithm
      [16-22)keVPSNR /dB24.851628.743438.886439.7786
      SSIM0.28210.55460.96870.9708
      RMSE0.02820.01690.00570.0055
      [28-31)keVPSNR /dB24.314328.617938.750939.4527
      SSIM0.26510.54540.96670.9719
      RMSE0.01050.00610.00210.0018
      [41-50)keVPSNR /dB22.421926.677532.223834.1206
      SSIM0.21490.45750.94340.9482
      RMSE0.00580.00330.00170.0015
    • Table 3. Quantitative results of reconstructed images obtained by different algorithms on the simulated data under noise interference

      View table

      Table 3. Quantitative results of reconstructed images obtained by different algorithms on the simulated data under noise interference

      ChannelParameterSARTASD-POCSSISTERProposed algorithm
      [16-22)keVPSNR /dB20.819228.727637.076637.1523
      SSIM0.14890.49570.93510.9360
      RMSE0.05250.01870.00780.0076
      [28-31)keVPSNR /dB19.905727.961936.957637.0246
      SSIM0.13920.47370.93510.9383
      RMSE0.01980.00690.00280.0027
      [41-50)keVPSNR /dB17.715424.802232.153932.7097
      SSIM0.11590.34920.89540.9011
      RMSE0.01140.00400.00200.0019
    • Table 4. Iteration time of different algorithms for different problems

      View table

      Table 4. Iteration time of different algorithms for different problems

      Problem nameNumber of iterationsIteration time of SISTERIteration time of the proposed algorithm
      Incomplete dataset10089.6 min21.7 min
      153.8 s13.0 s
      Noise interference100166.9 min44.7 min
      1100.1 s26.8 s
    Tools

    Get Citation

    Copy Citation Text

    Yizhong Wang, Ailong Cai, Ningning Liang, Xiaohuan Yu, Junru Ren, Lei Li, Bin Yan. Dual Plug and Play Multi-Energy Computed Tomography Reconstruction Algorithm[J]. Acta Optica Sinica, 2023, 43(14): 1434001

    Download Citation

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

    Category: X-Ray Optics

    Received: Feb. 22, 2023

    Accepted: Apr. 3, 2023

    Published Online: Jul. 13, 2023

    The Author Email: Bin Yan (ybspace@hotmail.com)

    DOI:10.3788/AOS230582

    Topics