• Acta Optica Sinica
  • Vol. 38, Issue 8, 0811003 (2018)
Haijiao Zhang1、2、*, Huihua Kong1、2、*, and Yonggang Sun1、2
Author Affiliations
  • 1 College of Science, North University of China, Taiyuan, Shanxi 0 30051, China
  • 2 Shanxi Key Laboratory of Signal Capturing & Processing, North University of China, Taiyuan, Shanxi 0 30051, China;
  • show less
    DOI: 10.3788/AOS201838.0811003 Cite this Article Set citation alerts
    Haijiao Zhang, Huihua Kong, Yonggang Sun. Weighted NLTV Reconstruction Algorithm Based on Structural Prior Information for Spectral CT[J]. Acta Optica Sinica, 2018, 38(8): 0811003 Copy Citation Text show less
    Three transformations apply to head model. (a) TV operator; (b) NLTV operator; (c) RE_NLTV operator
    Fig. 1. Three transformations apply to head model. (a) TV operator; (b) NLTV operator; (c) RE_NLTV operator
    Energy spectrum of 120 kVp simulated voltages
    Fig. 2. Energy spectrum of 120 kVp simulated voltages
    Two-dimensional image of the simulation model
    Fig. 3. Two-dimensional image of the simulation model
    Reference images used in RI_NLTV algorithm. (a) For chapter 3.1; (b) for chapter 3.2
    Fig. 4. Reference images used in RI_NLTV algorithm. (a) For chapter 3.1; (b) for chapter 3.2
    Reconstructions with different algorithms under high and low noise level. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    Fig. 5. Reconstructions with different algorithms under high and low noise level. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    Convergence of different algorithms under the low noise level
    Fig. 6. Convergence of different algorithms under the low noise level
    Reconstructions with different algorithms underthree energy bins. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    Fig. 7. Reconstructions with different algorithms underthree energy bins. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    Reconstructions with different algorithms under high and low noise level in the mouse model. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    Fig. 8. Reconstructions with different algorithms under high and low noise level in the mouse model. (a) TV algorithm; (b) NLTV algorithm; (c) RE_NLTV algorithm; (d) RI_NLTV algorithm
    ObjectCenter /cmSize /cmMaterial
    1(0.00,0.00)(0.92,0.69)Soft tissue
    2(0.20,0.20)(0.03,0.03)Air
    3(0.00,-0.10)(0.05,0.05)Blood
    4(0.00,0.50)(0.20,0.08)0.3% Iodine+99.7% blood
    5(0.65,0.20)(0.20,0.08)0.2% Kalium+99.8% blood
    6(0.45,-0.38)(0.20,0.08)1% Barium+99% water
    7(-0.45,-0.38)(0.20,0.08)0.3% Ferrum+99.7% blood
    8(-0.65,0.20)(0.20,0.08)10% Calcium+90% water
    Table 1. Position and size of the corresponding material in the model
    ItemLow noise levelHigh noise level
    TVNLTVRE_NLTVRI_NLTVTVNLTVRE_NLTVRI_NLTV
    SNR19.384518.166718.408621.118317.314617.110017.201219.9054
    MSE8.62×10-51.14×10-41.07×10-45.78×10-51.39×10-41.46×10-41.42×10-47.64×10-4
    NMSD0.10730.12350.12010.08790.13620.13950.13800.1011
    Table 2. Evaluation parameters of reconstruction with different algorithms under high and low noise level
    AlgorithmBin1Bin2Bin3
    SNRMSENMSDSNRMSENMSDSNRMSENMSD
    TV17.400.00110.134817.712.62×10-40.130217.643.06×10-40.1312
    NLTV16.570.00140.148416.803.22×10-40.144416.793.73×10-40.1448
    RE_NLTV17.060.00120.140317.123.00×10-40.139317.093.48×10-40.1398
    RI_NLTV17.640.00110.131218.192.70×10-40.123218.372.25×10-40.1207
    Table 3. Evaluation parameter of different algorithms under three energy bins
    ItemLow noise levelHigh noise level
    TVNLTVRE_NLTVRI_NLTVTVNLTVRE_NLTVRI_NLTV
    SNR16.452716.522116.607317.290016.430916.380316.457217.3558
    MSE6.17×10-45.20×10-45.10×10-44.36×10-55.31×10-45.37×10-45.28×10-44.29×10-4
    NMSD0.15040.14920.14780.13660.15080.15170.15040.1356
    Table 4. Evaluation parameters of reconstruction with different algorithms under high and low noise level
    Haijiao Zhang, Huihua Kong, Yonggang Sun. Weighted NLTV Reconstruction Algorithm Based on Structural Prior Information for Spectral CT[J]. Acta Optica Sinica, 2018, 38(8): 0811003
    Download Citation