• Optics and Precision Engineering
  • Vol. 29, Issue 11, 2529 (2021)
Xing ZHAO1,2,*, Tong-jun LIU1, Ping CHEN1, Jing-fan WANG1, and Wei-wei LIU1,2
Author Affiliations
  • 1Institute of Modern Optics, Nankai University, Tianjin300350, China
  • 2Tianjin Key Laboratory of Micro-scale Optical Information Science and Technology, Tianjin300350, China
  • show less
    DOI: 10.37188/OPE.2021.0219 Cite this Article
    Xing ZHAO, Tong-jun LIU, Ping CHEN, Jing-fan WANG, Wei-wei LIU. Extending dynamic range of detector in non-contact diffuse optical tomography system using deep learning[J]. Optics and Precision Engineering, 2021, 29(11): 2529 Copy Citation Text show less
    Measurement process of diffuse optical tomography
    Fig. 1. Measurement process of diffuse optical tomography
    Flow chart of diffuse optical tomography reconstruction algorithm
    Fig. 2. Flow chart of diffuse optical tomography reconstruction algorithm
    Schematic diagram of fully connected network
    Fig. 3. Schematic diagram of fully connected network
    Errors of training network with different training sets
    Fig. 4. Errors of training network with different training sets
    Recovery results of truncated data of different samples
    Fig. 5. Recovery results of truncated data of different samples
    Reconstructed absorption coefficient distribution of truncated data
    Fig. 6. Reconstructed absorption coefficient distribution of truncated data
    Comparison of reconstructed absorption coefficient distributions of different samples
    Fig. 7. Comparison of reconstructed absorption coefficient distributions of different samples
    Experimental optical path and distribution of phantom parameter
    Fig. 8. Experimental optical path and distribution of phantom parameter
    Measured data and data after network recovery
    Fig. 9. Measured data and data after network recovery
    Comparison of DOT reconstruction results of experimental data and its predicted values
    Fig. 10. Comparison of DOT reconstruction results of experimental data and its predicted values
    测试样本号训练集样本量1 000的网络训练集样本量4 000的网络
    RMSEPCCRMSEPCC
    10.018 30.628 60.017 20.701 7
    20.047 00.616 40.042 70.695 1
    30.054 80.654 20.051 00.714 8
    Table 1. Quantitative evaluation results of absorption coefficient of recovery data of network trained by different training sets
    Xing ZHAO, Tong-jun LIU, Ping CHEN, Jing-fan WANG, Wei-wei LIU. Extending dynamic range of detector in non-contact diffuse optical tomography system using deep learning[J]. Optics and Precision Engineering, 2021, 29(11): 2529
    Download Citation