• Optical Instruments
  • Vol. 45, Issue 4, 24 (2023)
Han YANG, Baicheng LI*, and Lingling CHEN*
Author Affiliations
  • School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, Shanghai 200093
  • show less
    DOI: 10.3969/j.issn.1005-5630.2023.004.004 Cite this Article
    Han YANG, Baicheng LI, Lingling CHEN. Improved Res-UNet-based vascular segmentation of retinal images[J]. Optical Instruments, 2023, 45(4): 24 Copy Citation Text show less
    Schematic diagram of the residual module
    Fig. 1. Schematic diagram of the residual module
    Schematic diagram of the nonlocal attention mechanism
    Fig. 2. Schematic diagram of the nonlocal attention mechanism
    Improved residual U-net
    Fig. 3. Improved residual U-net
    Comparison of segmentation results
    Fig. 4. Comparison of segmentation results
    Segmentation of partial vascular area
    Fig. 5. Segmentation of partial vascular area
    AlgorithmAccSeSpDice系数
    A10.95280.76560.98040.8170
    A20.95620.80150.98320.8197
    A30.95860.81960.98610.8249
    A40.96790.82450.98960.8281
    Table 1. Comparison of results from different structural segmentation methods based on U-Net
    AlgorithmAccSeSpDice系数
    method in Ref. [20] 0.95310.76420.97570.8142
    method in Ref. [21] 0.95330.79950.95330.8208
    method in Ref. [22] 0.96200.77630.97680.8172
    method in Ref. [23] 0.95460.74200.98200.8237
    method in Ref. [24] 0.94430.80530.97050.8155
    method in Ref. [25] 0.97670.81730.97330.8269
    method in Ref. [26] 0.96140.80950.97690.8117
    method in Ref. [27] 0.96380.81260.98140.8202
    Our method0.96790.82450.98960.8281
    Table 2. Comparison of results from different algorithms on DRIVE dataset
    Han YANG, Baicheng LI, Lingling CHEN. Improved Res-UNet-based vascular segmentation of retinal images[J]. Optical Instruments, 2023, 45(4): 24
    Download Citation