Opto-Electronic Engineering, Volume. 51, Issue 1, 230276-1(2024)

Retinal lesions graded algorithm that integrates coordinate perception and hybrid extraction

Liming Liang... Jiaxin Jin*, Yao Feng and Baohe Lu |Show fewer author(s)
Author Affiliations
  • School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou, Jiangxi 341000, China
  • show less
    Figures & Tables(11)
    The overall framework of the algorithm
    The hybrid dual model
    The multi -layer coordinate perception module
    Different DR hierarchical images pre -processing comparison. (a) Primitive images; (b) Pre-processing images
    The training loss curves of the proposed algorithm on (a) the IDRID dataset and (b) the APTOS 2019 dataset
    Network feature hot pictures. (a) Initial images; (b) Thermal images
    The confused matrix
    Reapped various types of AUC values. (a) ResNet-50+FDM [22]; (b) ResNet-50[23]; (c) Method of this article
    • Table 1. The ablation results of the IDRID dataset

      View table
      View in Article

      Table 1. The ablation results of the IDRID dataset

      ModelAcc/%QWK/%Se/%Sp/%
      M178.6488.3094.2094.11
      M278.6487.1391.3094.11
      M379.6186.1394.2091.17
      M480.5888.5486.9597.05
      M579.6188.4491.3097.05
      M681.5588.7694.2097.05
    • Table 2. The results of different algorithms in IDRID data sets

      View table
      View in Article

      Table 2. The results of different algorithms in IDRID data sets

      MethodsModelQWK/%Acc/%Se/%Sp/%
      注:Re和Se计算公式一致,能够相互替换。
      Shi et al.[7]Efficientnet-b587.6379.06
      Bhardwaj et al.[19]ResNet-5079.4682.8576.98
      Wu et al.[20]ResNet-1856.1964.2187.39
      Liang et al.[21]ResNet-50+FCFM88.7080.5894.2094.10
      Song et al.[22]ResNet-50+FDM89.1180.5892.7594.10
      Liu et al.[23]ResNet-5084.4275.7892.7591.17
      OursRes2Net-50+ Densenet-12188.7681.5594.2097.05
    • Table 3. The results of different models in the APTOS 2019 data sets

      View table
      View in Article

      Table 3. The results of different models in the APTOS 2019 data sets

      MethodsModelQWK/%Acc/%Re/%AUC/%
      注:Re和Se计算公式一致,能够相互替换。
      Shaik et al.[8]LA-NSVM75.6484.3166.16
      Bodapati et al.[24]Xception+VGG1682.5483.0079.00
      Bodapati et al.[25]VGG16-fc2+Xception70.9080.96
      Kobat et al.[26]DenseNet20178.3785.9369.72
      Song et al.[22]ResNet-50+FDM86.3484.2885.3292.78
      Liu et al.[23]ResNet-5086.0881.9686.4392.46
      OursRes2Net-50+ Densenet-12190.2984.4287.4093.60
    Tools

    Get Citation

    Copy Citation Text

    Liming Liang, Jiaxin Jin, Yao Feng, Baohe Lu. Retinal lesions graded algorithm that integrates coordinate perception and hybrid extraction[J]. Opto-Electronic Engineering, 2024, 51(1): 230276-1

    Download Citation

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

    Category: Article

    Received: Nov. 15, 2023

    Accepted: Jan. 4, 2024

    Published Online: Apr. 19, 2024

    The Author Email: Jin Jiaxin (金家新)

    DOI:10.12086/oee.2024.230276

    Topics