Laser & Optoelectronics Progress, Volume. 60, Issue 2, 0217002(2023)

Gland and Colonoscopy Segmentation Method Combining Self-Attention and Convolutional Neural Network

Jiabao Zhang and Zhiyong Xiao*
Author Affiliations
  • School of Artificial Intelligence and Computer Science, Jiangnan University, Wuxi 214122, Jiangsu , China
  • show less
    Figures & Tables(9)
    Local UNet encoding structure
    Cross fusion block structure
    LG UNet structure diagram
    Comparison of output results of each network
    • Table 1. Results of ablation experiment

      View table

      Table 1. Results of ablation experiment

      ModelDiceIOU
      Test ATest BTest ATest B
      U-Net0.88260.82690.79340.7216
      Local UNet0.91330.87460.84470.7868
      LG-v10.92670.88220.86630.8010
      LG-v20.93020.88340.87110.8028
      LG-v30.93620.88440.88170.8049
    • Table 2. Double channel and single channel output results are compared

      View table

      Table 2. Double channel and single channel output results are compared

      ModelDiceIOU
      Test ATest BTest ATest B
      U-Net -20.88260.82690.79340.7216
      U-Net -10.88190.85390.79560.7606
      LG UNet -20.93620.88440.88170.8049
      LG UNet -10.91000.86630.83800.7858
    • Table 3. Comparison results with other algorithms on Glas dataset

      View table

      Table 3. Comparison results with other algorithms on Glas dataset

      ModelDiceIOU
      Test ATest BMeanTest ATest BMean
      U-Net0.88260.82690.85470.79340.72160.7575
      ResUNet0.89400.84590.86990.81460.74610.7803
      KiU-Net280.88980.85270.87120.80650.75650.7815
      MedT0.86740.80510.83620.77110.68890.7300
      Rota-Net290.9190.8490.884
      U-Node300.89300.84200.8675
      LG UNet0.93620.88440.91030.88170.80490.8433
    • Table 4. Comparison results with other algorithms on Kvasir-SEG dataset

      View table

      Table 4. Comparison results with other algorithms on Kvasir-SEG dataset

      ModelDiceIOU
      Test ATest A
      U-Net0.81100.7250
      KiU-Net280.77240.6689
      ResUNet0.83060.7477
      ResUNet++0.85080.7699
      ColonSegNet310.8206
      MedT0.80390.7116
      LG UNet0.85630.7782
    • Table 5. Comparisons on other data sets

      View table

      Table 5. Comparisons on other data sets

      ModelClinicDBNoMuSeg
      DiceIOUDiceIOU
      U-Net0.79260.71150.79300.7188
      LG UNet0.86070.79460.83170.7126
    Tools

    Get Citation

    Copy Citation Text

    Jiabao Zhang, Zhiyong Xiao. Gland and Colonoscopy Segmentation Method Combining Self-Attention and Convolutional Neural Network[J]. Laser & Optoelectronics Progress, 2023, 60(2): 0217002

    Download Citation

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

    Category: Medical Optics and Biotechnology

    Received: Oct. 9, 2021

    Accepted: Nov. 16, 2021

    Published Online: Jan. 6, 2023

    The Author Email: Zhiyong Xiao (zhiyong.xiao@jiangnan.edu.cn)

    DOI:10.3788/LOP212696

    Topics