• Opto-Electronic Engineering
  • Vol. 51, Issue 7, 240114 (2024)
Minghui Chen1,*, Yanqi Lu1, Wenyi Yang1, Yuanzhu Wang2, and Yi Shao3
Author Affiliations
  • 1Shanghai Engineering Research Center of Interventional Medical, Shanghai Institute for Interventional Medical Devices, School of Health Sciences and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
  • 2Shanghai Raykeen Laser Technology Co., Ltd., Shanghai 200120, China
  • 3Shanghai General Hospital, Shanghai 200080, China
  • show less
    DOI: 10.12086/oee.2024.240114 Cite this Article
    Minghui Chen, Yanqi Lu, Wenyi Yang, Yuanzhu Wang, Yi Shao. Super-resolution reconstruction of retinal OCT image using multi-teacher knowledge distillation network[J]. Opto-Electronic Engineering, 2024, 51(7): 240114 Copy Citation Text show less
    Overall framework of MK-OCT
    Fig. 1. Overall framework of MK-OCT
    Structure of PASRN
    Fig. 2. Structure of PASRN
    Structure of PANet
    Fig. 3. Structure of PANet
    ECD module
    Fig. 4. ECD module
    Contrastive learning
    Fig. 5. Contrastive learning
    Results of super-resolution reconstruction
    Fig. 6. Results of super-resolution reconstruction
    MethodSize /MBFLOPs /GDataset 1Dataset 2
    PSNRSSIMLPIPSPIPSNRSSIMLPIPSPI
    Bicubic--28.120.78110.4126.79528.430.77300.4226.579
    SRCNN0.20.2328.590.80030.4046.35528.790.79860.3986.297
    CSD12.16122.130.950.81420.3105.67730.900.81190.3275.802
    IMDN2.6541.931.240.82170.2265.55331.210.82200.2305.608
    RFDN1.5932.031.670.82620.2205.21731.780.82170.2175.139
    MK-OCT (Ours)1.4129.832.930.84600.1494.52132.900.84430.1434.443
    Table 1. Average performance of various super-resolution models after x4 reconstruction
    DatasetMetricHR-SMKSingle-teacherNone-CL
    TPSNRTPI
    Dataset 1PSNR31.2732.8832.7732.88
    SSIM0.82380.84590.83960.8457
    LPIPS0.2300.2170.1420.150
    PI5.5935.4404.4574.608
    Dataset 2PSNR31.3332.8732.8132.86
    SSIM0.81780.84240.84110.8420
    LPIPS0.2140.2090.1400.148
    PI5.1465.1294.5614.670
    Table 2. Quantitative evaluation of student networks under different conditions after x4 reconstruction
    MethodPSNRPI
    ×2×4×2×4
    SRCNN33.6728.794.6676.033
    CSD34.2229.984.1095.820
    IMDN35.9031.064.2335.709
    RFDN35.8931.774.0595.455
    MK-OCT (Ours)36.2032.583.9795.103
    Table 3. Average PSRN and PI values of various super-resolution models after reconstruction
    Minghui Chen, Yanqi Lu, Wenyi Yang, Yuanzhu Wang, Yi Shao. Super-resolution reconstruction of retinal OCT image using multi-teacher knowledge distillation network[J]. Opto-Electronic Engineering, 2024, 51(7): 240114
    Download Citation