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. 59, Issue 16, 1617002(2022)

Breast Mass Segmentation Based on U-Net++ and Adversarial Learning Network

Yuanzhi Xie1, Shiju Yan1、*, Gaofeng Wei2, and Linying Yang1
Author Affiliations
  • 1School of Medical Instrument and Food Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China
  • 2Institute of Tropical Medicine, Naval Medical University, Shanghai 200025, China
    • show less
    AbstractGet PDF(in Chinese)
    Figures&Tables (8)
    Equations (7)
    References (23)
    Cited By (1)
    Tools
    Paper Information
    Topics
    Figures & Tables(8)
    Concise segmentation process

    Fig. 1. Concise segmentation process

    Download full size
    Overall frame diagram corresponding to the segmentation of breast mass

    Fig. 2. Overall frame diagram corresponding to the segmentation of breast mass

    Download full size
    Main modules of the improved U-Net++ network. (a) Encoder structure unit; (b) decoder structure unit

    Fig. 3. Main modules of the improved U-Net++ network. (a) Encoder structure unit; (b) decoder structure unit

    Download full size
    Discriminator network framework diagram

    Fig. 4. Discriminator network framework diagram

    Download full size
    Adversarial training process of generator and discriminator

    Fig. 5. Adversarial training process of generator and discriminator

    Download full size
    ROC curves of four types of networks

    Fig. 6. ROC curves of four types of networks

    Download full size
    Segmentation results of a breast mass. (a) Region of the breast mass to be segmented; (b) gold standard; (c) mask image produced by U-Net; (d) mask image produced by U-Net++; (e) mask image generated by U-Net+++adv; (f) mask image generated by U-Net+++adv+batch normalization

    Fig. 7. Segmentation results of a breast mass. (a) Region of the breast mass to be segmented; (b) gold standard; (c) mask image produced by U-Net; (d) mask image produced by U-Net++; (e) mask image generated by U-Net+++adv; (f) mask image generated by U-Net+++adv+batch normalization

    Download full size

    • Table 1. Segmentation results of four types of networks for breast masses

      View table

      Table 1. Segmentation results of four types of networks for breast masses

      MethodSpecificitySensitivityAccuracyDiceAUC
      U-Net0.9680.8420.9570.8780.8561
      U-Net++0.9740.8640.9640.8900.8745
      U-Net+++adv0.9880.8800.9730.9060.9085
      Proposed method0.9970.9040.9800.9100.9362
    Tools

    Get Citation

    Copy Citation Text

    Yuanzhi Xie, Shiju Yan, Gaofeng Wei, Linying Yang. Breast Mass Segmentation Based on U-Net++ and Adversarial Learning Network[J]. Laser & Optoelectronics Progress, 2022, 59(16): 1617002

    Download Citation

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

    Category: Medical Optics and Biotechnology

    Received: Aug. 13, 2021

    Accepted: Aug. 30, 2021

    Published Online: Jul. 26, 2022

    The Author Email: Shiju Yan (yanshiju@usst.edu.cn)

    DOI:10.3788/LOP202259.1617002

    Topics

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