Journals Articles Conferences News About CLP
Submit a paper Email Alert
Search
Search
Articles
Journals
News
Advanced Search
Top Searches
2D MaterialsTransformation opticsQuantum PhotonicsSmall lasersSemiconductor UV PhotonicsSupersymmetric microring laser arrays

Laser & Optoelectronics Progress, Volume. 57, Issue 16, 161015(2020)

Detection of Pulmonary Nodules Based on Improved Full Convolution Network Model

Ruoyu Liu and Libo Liu*
Author Affiliations
  • School of Information Engineering, Ningxia University, Yinchuan, Ningxia 750021, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (11)
    Equations (0)
    References (25)
    Cited By (2)
    Tools
    Paper Information
    Topics
    Figures & Tables(11)
    Structure of YOLACT model

    Fig. 1. Structure of YOLACT model

    Download full size
    Forecast output comparison of YOLACT and RetinaNet

    Fig. 2. Forecast output comparison of YOLACT and RetinaNet

    Download full size
    Process of DetNet generating a feature graph

    Fig. 3. Process of DetNet generating a feature graph

    Download full size
    FPN structure of DetNet

    Fig. 4. FPN structure of DetNet

    Download full size
    Diagram of transfer learning process

    Fig. 5. Diagram of transfer learning process

    Download full size
    Diagram of RReLU activation function

    Fig. 6. Diagram of RReLU activation function

    Download full size
    Detection results of pulmonary nodules

    Fig. 7. Detection results of pulmonary nodules

    Download full size
    ROC diagram of experimental model

    Fig. 8. ROC diagram of experimental model

    Download full size

    • Table 1. Comparison of YOLACT model before and after improvement

      View table

      Table 1. Comparison of YOLACT model before and after improvement

      ModelBackbonePFPR/%PFNR/%PACC/%
      YOLACT(base)ResNet+FPN3.975.8792.01
      YOLACT(Det)DetNet+FPN3.754.8792.84
      YOLACT(Tran)DetNet+FPN+Transfer learning3.454.0793.21
      YOLAC(RReLU)DetNet+FPN+RReLU3.584.0193.15
      YOLACT(ours)DetNet+FPN+Transfer learning+RReLU3.073.6594.58

    • Table 2. Comparison of false positive rate of different pulmonary nodule algorithms

      View table

      Table 2. Comparison of false positive rate of different pulmonary nodule algorithms

      MethodNTP/888=0.125NTP/888=0.25NTP/888=0.5NTP/888=1NTP/888=2NTP/888=4NTP/888=8CPM
      Method in Ref.[8]0.6920.7100.8090.8630.8950.9140.9230.838
      Method in Ref.[9]0.5170.6020.7200.7880.8220.8390.8560.735
      Method in Ref.[11]0.6770.7370.8150.8480.8790.9070.9220.827
      YOLACT(base)0.7150.7220.7950.8570.8820.9070.9190.828
      YOLACT(ours)0.7200.7350.8250.8630.9010.9110.9320.841

    • Table 3. Comparison of different pulmonary nodule algorithms

      View table

      Table 3. Comparison of different pulmonary nodule algorithms

      AlgorithmPFPR /%PFNR /%PACC /%
      Algorithm in Ref.[5]4.886.5191.10
      Algorithm in Ref.[6]3.516.2191.75
      Algorithm in Ref.[7]3.155.8993.50
      YOLACT(base)3.975.8792.01
      YOLACT(ours)3.123.6594.58
    Tools

    Get Citation

    Copy Citation Text

    Ruoyu Liu, Libo Liu. Detection of Pulmonary Nodules Based on Improved Full Convolution Network Model[J]. Laser & Optoelectronics Progress, 2020, 57(16): 161015

    Download Citation

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

    Category: Image Processing

    Received: Nov. 13, 2019

    Accepted: Jan. 16, 2020

    Published Online: Aug. 5, 2020

    The Author Email: Libo Liu (liulib@163.com)

    DOI:10.3788/LOP57.161015

    Topics

    laser devices and laser physics
    Lasers and Laser Optics
    Laser physics
    laser manufacturing
    Instrumentation, Measurement and Metrology